Update documentation for sstables conversions

Co-authored-by: tgrabiec <283695+tgrabiec@users.noreply.github.com>

.github/CODEOWNERS

@@ -1,5 +1,5 @@
 # AUTH
-auth/* @nuivall
+auth/* @nuivall @ptrsmrn
 
 # CACHE
 row_cache* @tgrabiec
@@ -25,11 +25,11 @@ compaction/* @raphaelsc
 transport/*
 
 # CQL QUERY LANGUAGE
-cql3/* @tgrabiec @nuivall
+cql3/* @tgrabiec @nuivall @ptrsmrn
 
 # COUNTERS
-counters* @nuivall
-tests/counter_test* @nuivall
+counters* @nuivall @ptrsmrn
+tests/counter_test* @nuivall @ptrsmrn
 
 # DOCS
 docs/* @annastuchlik @tzach
@@ -57,6 +57,7 @@ repair/* @tgrabiec @asias
 
 # SCHEMA MANAGEMENT
 db/schema_tables* @tgrabiec
+db/legacy_schema_migrator* @tgrabiec
 service/migration* @tgrabiec
 schema* @tgrabiec
 

.github/scripts/auto-backport.py

@@ -62,7 +62,7 @@ def create_pull_request(repo, new_branch_name, base_branch_name, pr, backport_pr
         if is_draft:
             labels_to_add.append("conflicts")
             pr_comment = f"@{pr.user.login} - This PR was marked as draft because it has conflicts\n"
-            pr_comment += "Please resolve them and remove the 'conflicts' label. The PR will be made ready for review automatically."
+            pr_comment += "Please resolve them and mark this PR as ready for review"
             backport_pr.create_issue_comment(pr_comment)
         
         # Apply all labels at once if we have any

.github/workflows/backport-pr-fixes-validation.yaml

@@ -18,7 +18,7 @@ jobs:
             
             // Regular expression pattern to check for "Fixes" prefix
             // Adjusted to dynamically insert the repository full name
-            const pattern = `Fixes:? ((?:#|${repo.replace('/', '\\/')}#|https://github\\.com/${repo.replace('/', '\\/')}/issues/)(\\d+)|([A-Z]+-\\d+))`;
+            const pattern = `Fixes:? (?:#|${repo.replace('/', '\\/')}#|https://github\\.com/${repo.replace('/', '\\/')}/issues/)(\\d+)`;
             const regex = new RegExp(pattern);
             
             if (!regex.test(body)) {

.github/workflows/docs-validate-metrics.yml

@@ -7,7 +7,7 @@ on:
       - enterprise
     paths:
       - '**/*.cc'
-      - 'scripts/metrics-config.yml'
+      - 'scripts/metrics-config.yml' 
       - 'scripts/get_description.py'
       - 'docs/_ext/scylladb_metrics.py'
 
@@ -15,20 +15,20 @@ jobs:
   validate-metrics:
     runs-on: ubuntu-latest
     name: Check metrics documentation coverage
-
+    
     steps:
     - name: Checkout code
       uses: actions/checkout@v4
       with:
         submodules: true
-
+      
     - name: Set up Python
       uses: actions/setup-python@v6
       with:
         python-version: '3.10'
-
+        
     - name: Install dependencies
       run: pip install PyYAML
-
+        
     - name: Validate metrics
       run: python3 scripts/get_description.py --validate -c scripts/metrics-config.yml

.github/workflows/trigger-scylla-ci.yaml

@@ -3,13 +3,10 @@ name: Trigger Scylla CI Route
 on:
   issue_comment:
     types: [created]
-  pull_request_target:
-    types:
-      - unlabeled
 
 jobs:
   trigger-jenkins:
-    if: (github.event.comment.user.login != 'scylladbbot' && contains(github.event.comment.body, '@scylladbbot') && contains(github.event.comment.body, 'trigger-ci')) || github.event.label.name == 'conflicts'
+    if: github.event.comment.user.login != 'scylladbbot' && contains(github.event.comment.body, '@scylladbbot') && contains(github.event.comment.body, 'trigger-ci')
     runs-on: ubuntu-latest
     steps:
       - name: Trigger Scylla-CI-Route Jenkins Job

SCYLLA_ASSERT_CONVERSION_SUMMARY.md

@@ -0,0 +1,182 @@
+# SCYLLA_ASSERT to scylla_assert() Conversion Summary
+
+## Objective
+
+Replace crash-inducing `SCYLLA_ASSERT` with exception-throwing `scylla_assert()` to prevent cluster-wide crashes and maintain availability.
+
+## What Was Done
+
+### 1. Infrastructure Implementation ✓
+
+Created new `scylla_assert()` macro in `utils/assert.hh`:
+- Based on `on_internal_error()` for exception-based error handling
+- Supports optional custom error messages via variadic arguments
+- Uses `seastar::format()` for string formatting
+- Compatible with C++23 standard (uses `__VA_OPT__`)
+
+**Key difference from SCYLLA_ASSERT:**
+```cpp
+// Old: Crashes the process immediately
+SCYLLA_ASSERT(condition);
+
+// New: Throws exception (or aborts based on config)
+scylla_assert(condition);
+scylla_assert(condition, "custom error message: {}", value);
+```
+
+### 2. Comprehensive Analysis ✓
+
+Analyzed entire codebase to identify safe vs unsafe conversion locations:
+
+**Statistics:**
+- Total SCYLLA_ASSERT usages: ~1307 (including tests)
+- Non-test usages: ~886
+- **Unsafe to convert**: 223 usages (25%)
+  - In noexcept functions: 187 usages across 50 files
+  - In destructors: 36 usages across 25 files
+- **Safe to convert**: ~668 usages (75%)
+- **Converted in this PR**: 112 usages (16.8% of safe conversions)
+
+### 3. Documentation ✓
+
+Created comprehensive documentation:
+
+1. **Conversion Guide** (`docs/dev/scylla_assert_conversion.md`)
+   - Explains safe vs unsafe contexts
+   - Provides conversion strategy
+   - Lists all completed conversions
+   - Includes testing guidance
+
+2. **Unsafe Locations Report** (`docs/dev/unsafe_scylla_assert_locations.md`)
+   - Detailed listing of 223 unsafe locations
+   - Organized by file with line numbers
+   - Separated into noexcept and destructor categories
+
+### 4. Sample Conversions ✓
+
+Converted 112 safe SCYLLA_ASSERT usages across 32 files as demonstration:
+
+| File | Conversions | Context |
+|------|------------|---------|
+| db/large_data_handler.{cc,hh} | 5 | Future-returning functions |
+| db/schema_applier.cc | 1 | Coroutine function |
+| db/system_distributed_keyspace.cc | 1 | Regular function |
+| db/commitlog/commitlog_replayer.cc | 1 | Coroutine function |
+| db/view/row_locking.cc | 2 | Regular function |
+| db/size_estimates_virtual_reader.cc | 1 | Lambda in coroutine |
+| db/corrupt_data_handler.cc | 2 | Lambdas in future-returning function |
+| raft/tracker.cc | 2 | Unreachable code (switch defaults) |
+| service/topology_coordinator.cc | 11 | Coroutine functions (topology operations) |
+| service/storage_service.cc | 28 | Critical node lifecycle operations |
+| sstables/* (22 files) | 58 | SSTable operations (read/write/compress/index) |
+
+All conversions were in **safe contexts** (non-noexcept, non-destructor functions). 3 assertions in storage_service.cc remain as SCYLLA_ASSERT (in noexcept functions).
+
+## Why These Cannot Be Converted
+
+### Unsafe Context #1: noexcept Functions (187 usages)
+
+**Problem**: Throwing from noexcept causes `std::terminate()`, same as crash.
+
+**Example** (from `locator/production_snitch_base.hh`):
+```cpp
+virtual bool prefer_local() const noexcept override {
+    SCYLLA_ASSERT(_backreference != nullptr);  // Cannot convert!
+    return _backreference->prefer_local();
+}
+```
+
+**Solution for these**: Keep as SCYLLA_ASSERT or use `on_fatal_internal_error()`.
+
+### Unsafe Context #2: Destructors (36 usages)
+
+**Problem**: Destructors are implicitly noexcept, throwing causes `std::terminate()`.
+
+**Example** (from `utils/file_lock.cc`):
+```cpp
+~file_lock() noexcept {
+    if (_fd.get() != -1) {
+        SCYLLA_ASSERT(_fd.get() != -1);  // Cannot convert!
+        auto r = ::flock(_fd.get(), LOCK_UN);
+        SCYLLA_ASSERT(r == 0);  // Cannot convert!
+    }
+}
+```
+
+**Solution for these**: Keep as SCYLLA_ASSERT.
+
+## Benefits of scylla_assert()
+
+1. **Prevents Cluster-Wide Crashes**
+   - Exception can be caught and handled gracefully
+   - Failed node doesn't bring down entire cluster
+
+2. **Maintains Availability**
+   - Service can continue with degraded functionality
+   - Better than complete crash
+
+3. **Better Error Reporting**
+   - Includes backtrace via `on_internal_error()`
+   - Supports custom error messages
+   - Configurable abort-on-error for testing
+
+4. **Backward Compatible**
+   - SCYLLA_ASSERT still exists for unsafe contexts
+   - Can be gradually adopted
+
+## Testing
+
+- Created manual test in `test/manual/test_scylla_assert.cc`
+- Verifies passing and failing assertions
+- Tests custom error messages
+- Code review passed with improvements made
+
+## Next Steps (Future Work)
+
+1. **Gradual Conversion**
+   - Convert remaining ~653 safe SCYLLA_ASSERT usages incrementally
+   - Prioritize high-impact code paths first
+
+2. **Review noexcept Functions**
+   - Evaluate if some can be made non-noexcept
+   - Consider using `on_fatal_internal_error()` where appropriate
+
+3. **Integration Testing**
+   - Run full test suite with conversions
+   - Monitor for any unexpected behavior
+   - Validate exception propagation
+
+4. **Automated Analysis Tool**
+   - Create tool to identify safe conversion candidates
+   - Generate conversion patches automatically
+   - Track conversion progress
+
+## Files Modified in This PR
+
+### Core Implementation
+- `utils/assert.hh` - Added scylla_assert() macro
+
+### Conversions
+- `db/large_data_handler.cc`
+- `db/large_data_handler.hh`
+- `db/schema_applier.cc`
+- `db/system_distributed_keyspace.cc`
+- `db/commitlog/commitlog_replayer.cc`
+- `db/view/row_locking.cc`
+- `db/size_estimates_virtual_reader.cc`
+- `db/corrupt_data_handler.cc`
+- `raft/tracker.cc`
+- `service/topology_coordinator.cc`
+- `service/storage_service.cc`
+- `sstables/` (22 files across trie/, mx/, and core sstables)
+
+### Documentation
+- `docs/dev/scylla_assert_conversion.md`
+- `docs/dev/unsafe_scylla_assert_locations.md`
+- `test/manual/test_scylla_assert.cc`
+
+## Conclusion
+
+This PR establishes the infrastructure and methodology for replacing SCYLLA_ASSERT with scylla_assert() to improve cluster availability. The sample conversions demonstrate the approach, while comprehensive documentation enables future work.
+
+**Key Achievement**: Provided a safe path forward for converting 75% (~668) of SCYLLA_ASSERT usages to exception-based assertions, while clearly documenting the 25% (~223) that must remain as crash-inducing assertions due to language constraints. Converted 112 usages as demonstration (16.8% of safe conversions), prioritizing critical files like storage_service.cc (node lifecycle) and all sstables files (data persistence), with ~556 remaining.

alternator/conditions.cc

@@ -42,7 +42,7 @@ comparison_operator_type get_comparison_operator(const rjson::value& comparison_
     if (!comparison_operator.IsString()) {
         throw api_error::validation(fmt::format("Invalid comparison operator definition {}", rjson::print(comparison_operator)));
     }
-    std::string op = rjson::to_string(comparison_operator);
+    std::string op = comparison_operator.GetString();
     auto it = ops.find(op);
     if (it == ops.end()) {
         throw api_error::validation(fmt::format("Unsupported comparison operator {}", op));
@@ -377,8 +377,8 @@ bool check_compare(const rjson::value* v1, const rjson::value& v2, const Compara
         return cmp(unwrap_number(*v1, cmp.diagnostic), unwrap_number(v2, cmp.diagnostic));
     }
     if (kv1.name == "S") {
-        return cmp(rjson::to_string_view(kv1.value),
-                   rjson::to_string_view(kv2.value));
+        return cmp(std::string_view(kv1.value.GetString(), kv1.value.GetStringLength()),
+                   std::string_view(kv2.value.GetString(), kv2.value.GetStringLength()));
     }
     if (kv1.name == "B") {
         auto d_kv1 = unwrap_bytes(kv1.value, v1_from_query);
@@ -470,9 +470,9 @@ static bool check_BETWEEN(const rjson::value* v, const rjson::value& lb, const r
         return check_BETWEEN(unwrap_number(*v, diag), unwrap_number(lb, diag), unwrap_number(ub, diag), bounds_from_query);
     }
     if (kv_v.name == "S") {
-        return check_BETWEEN(rjson::to_string_view(kv_v.value),
-                             rjson::to_string_view(kv_lb.value),
-                             rjson::to_string_view(kv_ub.value),
+        return check_BETWEEN(std::string_view(kv_v.value.GetString(), kv_v.value.GetStringLength()),
+                             std::string_view(kv_lb.value.GetString(), kv_lb.value.GetStringLength()),
+                             std::string_view(kv_ub.value.GetString(), kv_ub.value.GetStringLength()),
                              bounds_from_query);
     }
     if (kv_v.name == "B") {

alternator/consumed_capacity.cc

@@ -8,8 +8,6 @@
 
 #include "consumed_capacity.hh"
 #include "error.hh"
-#include "utils/rjson.hh"
-#include <fmt/format.h>
 
 namespace alternator {
 
@@ -34,12 +32,12 @@ bool consumed_capacity_counter::should_add_capacity(const rjson::value& request)
     if (!return_consumed->IsString()) {
         throw api_error::validation("Non-string ReturnConsumedCapacity field in request");
     }
-    std::string_view consumed = rjson::to_string_view(*return_consumed);
+    std::string consumed = return_consumed->GetString();
     if (consumed == "INDEXES") {
         throw api_error::validation("INDEXES consumed capacity is not supported");
     }
     if (consumed != "TOTAL") {
-        throw api_error::validation(fmt::format("Unknown consumed capacity {}", consumed));
+        throw api_error::validation("Unknown consumed capacity "+ consumed);
     }
     return true;
 }

alternator/executor.cc

@@ -419,7 +419,7 @@ static std::optional<std::string> find_table_name(const rjson::value& request) {
     if (!table_name_value->IsString()) {
         throw api_error::validation("Non-string TableName field in request");
     }
-    std::string table_name = rjson::to_string(*table_name_value);
+    std::string table_name = table_name_value->GetString();
     return table_name;
 }
 
@@ -546,7 +546,7 @@ get_table_or_view(service::storage_proxy& proxy, const rjson::value& request) {
             // does exist but the index does not (ValidationException).
             if (proxy.data_dictionary().has_schema(keyspace_name, orig_table_name)) {
                 throw api_error::validation(
-                    fmt::format("Requested resource not found: Index '{}' for table '{}'", rjson::to_string_view(*index_name), orig_table_name));
+                    fmt::format("Requested resource not found: Index '{}' for table '{}'", index_name->GetString(), orig_table_name));
             } else {
                 throw api_error::resource_not_found(
                     fmt::format("Requested resource not found: Table: {} not found", orig_table_name));
@@ -587,7 +587,7 @@ static std::string get_string_attribute(const rjson::value& value, std::string_v
         throw api_error::validation(fmt::format("Expected string value for attribute {}, got: {}",
                 attribute_name, value));
     }
-    return rjson::to_string(*attribute_value);
+    return std::string(attribute_value->GetString(), attribute_value->GetStringLength());
 }
 
 // Convenience function for getting the value of a boolean attribute, or a
@@ -1080,8 +1080,8 @@ static void add_column(schema_builder& builder, const std::string& name, const r
     }
     for (auto it = attribute_definitions.Begin(); it != attribute_definitions.End(); ++it) {
         const rjson::value& attribute_info = *it;
-        if (rjson::to_string_view(attribute_info["AttributeName"]) == name) {
-            std::string_view type = rjson::to_string_view(attribute_info["AttributeType"]);
+        if (attribute_info["AttributeName"].GetString() == name) {
+            auto type = attribute_info["AttributeType"].GetString();
             data_type dt = parse_key_type(type);
             if (computed_column) {
                 // Computed column for GSI (doesn't choose a real column as-is
@@ -1116,7 +1116,7 @@ static std::pair<std::string, std::string> parse_key_schema(const rjson::value&
         throw api_error::validation("First element of KeySchema must be an object");
     }
     const rjson::value *v = rjson::find((*key_schema)[0], "KeyType");
-    if (!v || !v->IsString() || rjson::to_string_view(*v) != "HASH") {
+    if (!v || !v->IsString() || v->GetString() != std::string("HASH")) {
         throw api_error::validation("First key in KeySchema must be a HASH key");
     }
     v = rjson::find((*key_schema)[0], "AttributeName");
@@ -1124,14 +1124,14 @@ static std::pair<std::string, std::string> parse_key_schema(const rjson::value&
         throw api_error::validation("First key in KeySchema must have string AttributeName");
     }
     validate_attr_name_length(supplementary_context, v->GetStringLength(), true, "HASH key in KeySchema - ");
-    std::string hash_key = rjson::to_string(*v);
+    std::string hash_key = v->GetString();
     std::string range_key;
     if (key_schema->Size() == 2) {
         if (!(*key_schema)[1].IsObject()) {
             throw api_error::validation("Second element of KeySchema must be an object");
         }
         v = rjson::find((*key_schema)[1], "KeyType");
-        if (!v || !v->IsString() || rjson::to_string_view(*v) != "RANGE") {
+        if (!v || !v->IsString() || v->GetString() != std::string("RANGE")) {
             throw api_error::validation("Second key in KeySchema must be a RANGE key");
         }
         v = rjson::find((*key_schema)[1], "AttributeName");
@@ -1887,8 +1887,8 @@ future<executor::request_return_type> executor::create_table(client_state& clien
         std::string def_type = type_to_string(def.type);
         for (auto it = attribute_definitions.Begin(); it != attribute_definitions.End(); ++it) {
             const rjson::value& attribute_info = *it;
-            if (rjson::to_string_view(attribute_info["AttributeName"]) == def.name_as_text()) {
-                std::string_view type = rjson::to_string_view(attribute_info["AttributeType"]);
+            if (attribute_info["AttributeName"].GetString() == def.name_as_text()) {
+                auto type = attribute_info["AttributeType"].GetString();
                 if (type != def_type) {
                     throw api_error::validation(fmt::format("AttributeDefinitions redefined {} to {} already a key attribute of type {} in this table", def.name_as_text(), type, def_type));
                 }
@@ -2223,12 +2223,12 @@ void validate_value(const rjson::value& v, const char* caller) {
 
 // The put_or_delete_item class builds the mutations needed by the PutItem and
 // DeleteItem operations - either as stand-alone commands or part of a list
-// of commands in BatchWriteItem.
+// of commands in BatchWriteItems.
 // put_or_delete_item splits each operation into two stages: Constructing the
 // object parses and validates the user input (throwing exceptions if there
 // are input errors). Later, build() generates the actual mutation, with a
 // specified timestamp. This split is needed because of the peculiar needs of
-// BatchWriteItem and LWT. BatchWriteItem needs all parsing to happen before
+// BatchWriteItems and LWT. BatchWriteItems needs all parsing to happen before
 // any writing happens (if one of the commands has an error, none of the
 // writes should be done). LWT makes it impossible for the parse step to
 // generate "mutation" objects, because the timestamp still isn't known.
@@ -2362,7 +2362,7 @@ put_or_delete_item::put_or_delete_item(const rjson::value& item, schema_ptr sche
     _cells = std::vector<cell>();
     _cells->reserve(item.MemberCount());
     for (auto it = item.MemberBegin(); it != item.MemberEnd(); ++it) {
-        bytes column_name = to_bytes(rjson::to_string_view(it->name));
+        bytes column_name = to_bytes(it->name.GetString());
         validate_value(it->value, "PutItem");
         const column_definition* cdef = find_attribute(*schema, column_name);
         validate_attr_name_length("", column_name.size(), cdef && cdef->is_primary_key());
@@ -2739,7 +2739,7 @@ future<executor::request_return_type> rmw_operation::execute(service::storage_pr
     auto read_command = needs_read_before_write ?
             previous_item_read_command(proxy, schema(), _ck, selection) :
             nullptr;
-    return proxy.cas(schema(), std::move(*cas_shard), *this, read_command, to_partition_ranges(*schema(), _pk),
+    return proxy.cas(schema(), std::move(*cas_shard), shared_from_this(), read_command, to_partition_ranges(*schema(), _pk),
             {timeout, std::move(permit), client_state, trace_state},
             db::consistency_level::LOCAL_SERIAL, db::consistency_level::LOCAL_QUORUM, timeout, timeout, true, std::move(cdc_opts)).then([this, read_command, &wcu_total] (bool is_applied) mutable {
         if (!is_applied) {
@@ -2783,10 +2783,10 @@ static void verify_all_are_used(const rjson::value* field,
         return;
     }
     for (auto it = field->MemberBegin(); it != field->MemberEnd(); ++it) {
-        if (!used.contains(rjson::to_string(it->name))) {
+        if (!used.contains(it->name.GetString())) {
             throw api_error::validation(
                 format("{} has spurious '{}', not used in {}",
-                    field_name, rjson::to_string_view(it->name), operation));
+                    field_name, it->name.GetString(), operation));
         }
     }
 }
@@ -3000,7 +3000,7 @@ future<executor::request_return_type> executor::delete_item(client_state& client
 }
 
 static schema_ptr get_table_from_batch_request(const service::storage_proxy& proxy, const rjson::value::ConstMemberIterator& batch_request) {
-    sstring table_name = rjson::to_sstring(batch_request->name); // JSON keys are always strings
+    sstring table_name = batch_request->name.GetString(); // JSON keys are always strings
     try {
         return proxy.data_dictionary().find_schema(sstring(executor::KEYSPACE_NAME_PREFIX) + table_name, table_name);
     } catch(data_dictionary::no_such_column_family&) {
@@ -3026,20 +3026,17 @@ struct primary_key_equal {
 };
 
 // This is a cas_request subclass for applying given put_or_delete_items to
-// one partition using LWT as part as BatchWriteItem. This is a write-only
+// one partition using LWT as part as BatchWriteItems. This is a write-only
 // operation, not needing the previous value of the item (the mutation to be
 // done is known prior to starting the operation). Nevertheless, we want to
 // do this mutation via LWT to ensure that it is serialized with other LWT
 // mutations to the same partition.
-// 
-// The std::vector<put_or_delete_item> must remain alive until the
-// storage_proxy::cas() future is resolved.
 class put_or_delete_item_cas_request : public service::cas_request {
     schema_ptr schema;
-    const std::vector<put_or_delete_item>& _mutation_builders;
+    std::vector<put_or_delete_item> _mutation_builders;
 public:
-    put_or_delete_item_cas_request(schema_ptr s, const std::vector<put_or_delete_item>& b) :
-        schema(std::move(s)), _mutation_builders(b) { }
+    put_or_delete_item_cas_request(schema_ptr s, std::vector<put_or_delete_item>&& b) :
+        schema(std::move(s)), _mutation_builders(std::move(b)) { }
     virtual ~put_or_delete_item_cas_request() = default;
     virtual std::optional<mutation> apply(foreign_ptr<lw_shared_ptr<query::result>> qr, const query::partition_slice& slice, api::timestamp_type ts, cdc::per_request_options& cdc_opts) override {
         std::optional<mutation> ret;
@@ -3055,48 +3052,20 @@ public:
     }
 };
 
-future<> executor::cas_write(schema_ptr schema, service::cas_shard cas_shard, const dht::decorated_key& dk,
-        const std::vector<put_or_delete_item>& mutation_builders, service::client_state& client_state,
-        tracing::trace_state_ptr trace_state, service_permit permit)
-{
-    if (!cas_shard.this_shard()) {
-        _stats.shard_bounce_for_lwt++;
-        return container().invoke_on(cas_shard.shard(), _ssg,
-                    [cs = client_state.move_to_other_shard(),
-                    &mb = mutation_builders,
-                    &dk,
-                    ks = schema->ks_name(),
-                    cf = schema->cf_name(),
-                    gt = tracing::global_trace_state_ptr(trace_state),
-                    permit = std::move(permit)]
-                    (executor& self) mutable {
-            return do_with(cs.get(), [&mb, &dk, ks = std::move(ks), cf = std::move(cf),
-                                    trace_state = tracing::trace_state_ptr(gt), &self]
-                                    (service::client_state& client_state) mutable {
-                auto schema = self._proxy.data_dictionary().find_schema(ks, cf);
-                service::cas_shard cas_shard(*schema, dk.token());
-
-                //FIXME: Instead of passing empty_service_permit() to the background operation,
-                // the current permit's lifetime should be prolonged, so that it's destructed
-                // only after all background operations are finished as well.
-                return self.cas_write(schema, std::move(cas_shard), dk, mb, client_state, std::move(trace_state), empty_service_permit());
-            });
-        });
-    }
-
+static future<> cas_write(service::storage_proxy& proxy, schema_ptr schema, service::cas_shard cas_shard, dht::decorated_key dk, std::vector<put_or_delete_item>&& mutation_builders,
+        service::client_state& client_state, tracing::trace_state_ptr trace_state, service_permit permit) {
     auto timeout = executor::default_timeout();
-    auto op = std::make_unique<put_or_delete_item_cas_request>(schema, mutation_builders);
-    auto* op_ptr = op.get();
+    auto op = seastar::make_shared<put_or_delete_item_cas_request>(schema, std::move(mutation_builders));
     auto cdc_opts = cdc::per_request_options{
         .alternator = true,
         .alternator_streams_increased_compatibility =
-                schema->cdc_options().enabled() && _proxy.data_dictionary().get_config().alternator_streams_increased_compatibility(),
+                schema->cdc_options().enabled() && proxy.data_dictionary().get_config().alternator_streams_increased_compatibility(),
     };
-    return _proxy.cas(schema, std::move(cas_shard), *op_ptr, nullptr, to_partition_ranges(dk),
+    return proxy.cas(schema, std::move(cas_shard), op, nullptr, to_partition_ranges(dk),
             {timeout, std::move(permit), client_state, trace_state},
             db::consistency_level::LOCAL_SERIAL, db::consistency_level::LOCAL_QUORUM,
-            timeout, timeout, true, std::move(cdc_opts)).finally([op = std::move(op)]{}).discard_result();
-    // We discarded cas()'s future value ("is_applied") because BatchWriteItem
+            timeout, timeout, true, std::move(cdc_opts)).discard_result();
+    // We discarded cas()'s future value ("is_applied") because BatchWriteItems
     // does not need to support conditional updates.
 }
 
@@ -3118,11 +3087,13 @@ struct schema_decorated_key_equal {
 
 // FIXME: if we failed writing some of the mutations, need to return a list
 // of these failed mutations rather than fail the whole write (issue #5650).
-future<> executor::do_batch_write(
+static future<> do_batch_write(service::storage_proxy& proxy,
+        smp_service_group ssg,
         std::vector<std::pair<schema_ptr, put_or_delete_item>> mutation_builders,
         service::client_state& client_state,
         tracing::trace_state_ptr trace_state,
-        service_permit permit) {
+        service_permit permit,
+        stats& stats) {
     if (mutation_builders.empty()) {
         return make_ready_future<>();
     }
@@ -3144,7 +3115,7 @@ future<> executor::do_batch_write(
             mutations.push_back(b.second.build(b.first, now));
             any_cdc_enabled |= b.first->cdc_options().enabled();
         }
-        return _proxy.mutate(std::move(mutations),
+        return proxy.mutate(std::move(mutations),
                 db::consistency_level::LOCAL_QUORUM,
                 executor::default_timeout(),
                 trace_state,
@@ -3153,48 +3124,55 @@ future<> executor::do_batch_write(
                 false,
                 cdc::per_request_options{
                     .alternator = true,
-                    .alternator_streams_increased_compatibility = any_cdc_enabled && _proxy.data_dictionary().get_config().alternator_streams_increased_compatibility(),
+                    .alternator_streams_increased_compatibility = any_cdc_enabled && proxy.data_dictionary().get_config().alternator_streams_increased_compatibility(),
                 });
     } else {
         // Do the write via LWT:
         // Multiple mutations may be destined for the same partition, adding
         // or deleting different items of one partition. Join them together
         // because we can do them in one cas() call.
-        using map_type = std::unordered_map<schema_decorated_key, 
-            std::vector<put_or_delete_item>, 
-            schema_decorated_key_hash, 
-            schema_decorated_key_equal>;
-        auto key_builders = std::make_unique<map_type>(1, schema_decorated_key_hash{}, schema_decorated_key_equal{});
-        for (auto&& b : std::move(mutation_builders)) {
-            auto [it, added] = key_builders->try_emplace(schema_decorated_key {
-                .schema = b.first,
-                .dk = dht::decorate_key(*b.first, b.second.pk())
-            });
+        std::unordered_map<schema_decorated_key, std::vector<put_or_delete_item>, schema_decorated_key_hash, schema_decorated_key_equal>
+            key_builders(1, schema_decorated_key_hash{}, schema_decorated_key_equal{});
+        for (auto& b : mutation_builders) {
+            auto dk = dht::decorate_key(*b.first, b.second.pk());
+            auto [it, added] = key_builders.try_emplace(schema_decorated_key{b.first, dk});
             it->second.push_back(std::move(b.second));
         }
-        auto* key_builders_ptr = key_builders.get();
-        return parallel_for_each(*key_builders_ptr, [this, &client_state, trace_state, permit = std::move(permit)] (const auto& e) {
-            _stats.write_using_lwt++;
+        return parallel_for_each(std::move(key_builders), [&proxy, &client_state, &stats, trace_state, ssg, permit = std::move(permit)] (auto& e) {
+            stats.write_using_lwt++;
             auto desired_shard = service::cas_shard(*e.first.schema, e.first.dk.token());
-            auto s = e.first.schema;
+            if (desired_shard.this_shard()) {
+                return cas_write(proxy, e.first.schema, std::move(desired_shard), e.first.dk, std::move(e.second), client_state, trace_state, permit);
+            } else {
+                stats.shard_bounce_for_lwt++;
+                return proxy.container().invoke_on(desired_shard.shard(), ssg,
+                            [cs = client_state.move_to_other_shard(),
+                             mb = e.second,
+                             dk = e.first.dk,
+                             ks = e.first.schema->ks_name(),
+                             cf = e.first.schema->cf_name(),
+                             gt =  tracing::global_trace_state_ptr(trace_state),
+                             permit = std::move(permit)]
+                            (service::storage_proxy& proxy) mutable {
+                    return do_with(cs.get(), [&proxy, mb = std::move(mb), dk = std::move(dk), ks = std::move(ks), cf = std::move(cf),
+                                              trace_state = tracing::trace_state_ptr(gt)]
+                                              (service::client_state& client_state) mutable {
+                        auto schema = proxy.data_dictionary().find_schema(ks, cf);
 
-            static const auto* injection_name = "alternator_executor_batch_write_wait";
-            return utils::get_local_injector().inject(injection_name, [s = std::move(s)] (auto& handler) -> future<> {
-                const auto ks = handler.get("keyspace");
-                const auto cf = handler.get("table");
-                const auto shard = std::atoll(handler.get("shard")->data());
-                if (ks == s->ks_name() && cf == s->cf_name() && shard == this_shard_id()) {
-                    elogger.info("{}: hit", injection_name);
-                    co_await handler.wait_for_message(std::chrono::steady_clock::now() + std::chrono::minutes{5});
-                    elogger.info("{}: continue", injection_name);
-                }
-            }).then([&e, desired_shard = std::move(desired_shard),
-                 &client_state, trace_state = std::move(trace_state), permit = std::move(permit), this]() mutable
-            {
-                return cas_write(e.first.schema, std::move(desired_shard), e.first.dk,
-                    std::move(e.second), client_state, std::move(trace_state), std::move(permit));
-            });
-        }).finally([key_builders = std::move(key_builders)]{});
+                        // The desired_shard on the original shard remains alive for the duration
+                        // of cas_write on this shard and prevents any tablet operations.
+                        // However, we need a local instance of cas_shard on this shard
+                        // to pass it to sp::cas, so we just create a new one.
+                        service::cas_shard cas_shard(*schema, dk.token());
+
+                        //FIXME: Instead of passing empty_service_permit() to the background operation,
+                        // the current permit's lifetime should be prolonged, so that it's destructed
+                        // only after all background operations are finished as well.
+                        return cas_write(proxy, schema, std::move(cas_shard), dk, std::move(mb), client_state, std::move(trace_state), empty_service_permit());
+                    });
+                }).finally([desired_shard = std::move(desired_shard)]{});
+            }
+        });
     }
 }
 
@@ -3341,7 +3319,7 @@ future<executor::request_return_type> executor::batch_write_item(client_state& c
     _stats.wcu_total[stats::DELETE_ITEM] += wcu_delete_units;
     _stats.api_operations.batch_write_item_batch_total += total_items;
     _stats.api_operations.batch_write_item_histogram.add(total_items);
-    co_await do_batch_write(std::move(mutation_builders), client_state, trace_state, std::move(permit));
+    co_await do_batch_write(_proxy, _ssg, std::move(mutation_builders), client_state, trace_state, std::move(permit), _stats);
     // FIXME: Issue #5650: If we failed writing some of the updates,
     // need to return a list of these failed updates in UnprocessedItems
     // rather than fail the whole write (issue #5650).
@@ -3386,7 +3364,7 @@ static bool hierarchy_filter(rjson::value& val, const attribute_path_map_node<T>
         }
         rjson::value newv = rjson::empty_object();
         for (auto it = v.MemberBegin(); it != v.MemberEnd(); ++it) {
-            std::string attr = rjson::to_string(it->name);
+            std::string attr = it->name.GetString();
             auto x = members.find(attr);
             if (x != members.end()) {
                 if (x->second) {
@@ -3606,7 +3584,7 @@ static std::optional<attrs_to_get> calculate_attrs_to_get(const rjson::value& re
         const rjson::value& attributes_to_get = req["AttributesToGet"];
         attrs_to_get ret;
         for (auto it = attributes_to_get.Begin(); it != attributes_to_get.End(); ++it) {
-            attribute_path_map_add("AttributesToGet", ret, rjson::to_string(*it));
+            attribute_path_map_add("AttributesToGet", ret, it->GetString());
             validate_attr_name_length("AttributesToGet", it->GetStringLength(), false);
         }
         if (ret.empty()) {
@@ -4272,12 +4250,12 @@ inline void update_item_operation::apply_attribute_updates(const std::unique_ptr
         attribute_collector& modified_attrs, bool& any_updates, bool& any_deletes) const {
     for (auto it = _attribute_updates->MemberBegin(); it != _attribute_updates->MemberEnd(); ++it) {
         // Note that it.key() is the name of the column, *it is the operation
-        bytes column_name = to_bytes(rjson::to_string_view(it->name));
+        bytes column_name = to_bytes(it->name.GetString());
         const column_definition* cdef = _schema->get_column_definition(column_name);
         if (cdef && cdef->is_primary_key()) {
-            throw api_error::validation(format("UpdateItem cannot update key column {}", rjson::to_string_view(it->name)));
+            throw api_error::validation(format("UpdateItem cannot update key column {}", it->name.GetString()));
         }
-        std::string action = rjson::to_string((it->value)["Action"]);
+        std::string action = (it->value)["Action"].GetString();
         if (action == "DELETE") {
             // The DELETE operation can do two unrelated tasks. Without a
             // "Value" option, it is used to delete an attribute. With a
@@ -5474,7 +5452,7 @@ calculate_bounds_conditions(schema_ptr schema, const rjson::value& conditions) {
     std::vector<query::clustering_range> ck_bounds;
 
     for (auto it = conditions.MemberBegin(); it != conditions.MemberEnd(); ++it) {
-        sstring key = rjson::to_sstring(it->name);
+        std::string key = it->name.GetString();
         const rjson::value& condition = it->value;
 
         const rjson::value& comp_definition = rjson::get(condition, "ComparisonOperator");
@@ -5482,13 +5460,13 @@ calculate_bounds_conditions(schema_ptr schema, const rjson::value& conditions) {
 
         const column_definition& pk_cdef = schema->partition_key_columns().front();
         const column_definition* ck_cdef = schema->clustering_key_size() > 0 ? &schema->clustering_key_columns().front() : nullptr;
-        if (key == pk_cdef.name_as_text()) {
+        if (sstring(key) == pk_cdef.name_as_text()) {
             if (!partition_ranges.empty()) {
                 throw api_error::validation("Currently only a single restriction per key is allowed");
             }
             partition_ranges.push_back(calculate_pk_bound(schema, pk_cdef, comp_definition, attr_list));
         }
-        if (ck_cdef && key == ck_cdef->name_as_text()) {
+        if (ck_cdef && sstring(key) == ck_cdef->name_as_text()) {
             if (!ck_bounds.empty()) {
                 throw api_error::validation("Currently only a single restriction per key is allowed");
             }
@@ -5889,7 +5867,7 @@ future<executor::request_return_type> executor::list_tables(client_state& client
 
     rjson::value* exclusive_start_json = rjson::find(request, "ExclusiveStartTableName");
     rjson::value* limit_json = rjson::find(request, "Limit");
-    std::string exclusive_start = exclusive_start_json ? rjson::to_string(*exclusive_start_json) : "";
+    std::string exclusive_start = exclusive_start_json ? exclusive_start_json->GetString() : "";
     int limit = limit_json ? limit_json->GetInt() : 100;
     if (limit < 1 || limit > 100) {
         co_return api_error::validation("Limit must be greater than 0 and no greater than 100");

alternator/executor.hh

@@ -40,7 +40,6 @@ namespace cql3::selection {
 
 namespace service {
     class storage_proxy;
-    class cas_shard;
 }
 
 namespace cdc {
@@ -58,7 +57,6 @@ class schema_builder;
 namespace alternator {
 
 class rmw_operation;
-class put_or_delete_item;
 
 schema_ptr get_table(service::storage_proxy& proxy, const rjson::value& request);
 bool is_alternator_keyspace(const sstring& ks_name);
@@ -221,16 +219,6 @@ private:
 
     static void describe_key_schema(rjson::value& parent, const schema&, std::unordered_map<std::string,std::string> * = nullptr, const std::map<sstring, sstring> *tags = nullptr);
 
-    future<> do_batch_write(
-        std::vector<std::pair<schema_ptr, put_or_delete_item>> mutation_builders,
-        service::client_state& client_state,
-        tracing::trace_state_ptr trace_state,
-        service_permit permit);
-
-    future<> cas_write(schema_ptr schema, service::cas_shard cas_shard, const dht::decorated_key& dk,
-        const std::vector<put_or_delete_item>& mutation_builders, service::client_state& client_state,
-        tracing::trace_state_ptr trace_state, service_permit permit);
-
 public:
     static void describe_key_schema(rjson::value& parent, const schema& schema, std::unordered_map<std::string,std::string>&, const std::map<sstring, sstring> *tags = nullptr);
 

alternator/serialization.cc

@@ -496,7 +496,7 @@ const std::pair<std::string, const rjson::value*> unwrap_set(const rjson::value&
         return {"", nullptr};
     }
     auto it = v.MemberBegin();
-    const std::string it_key = rjson::to_string(it->name);
+    const std::string it_key = it->name.GetString();
     if (it_key != "SS" && it_key != "BS" && it_key != "NS") {
         return {std::move(it_key), nullptr};
     }

alternator/ttl.cc

@@ -93,7 +93,7 @@ future<executor::request_return_type> executor::update_time_to_live(client_state
     if (v->GetStringLength() < 1 || v->GetStringLength() > 255) {
         co_return api_error::validation("The length of AttributeName must be between 1 and 255");
     }
-    sstring attribute_name = rjson::to_sstring(*v);
+    sstring attribute_name(v->GetString(), v->GetStringLength());
 
     co_await verify_permission(_enforce_authorization, _warn_authorization, client_state, schema, auth::permission::ALTER, _stats);
     co_await db::modify_tags(_mm, schema->ks_name(), schema->cf_name(), [&](std::map<sstring, sstring>& tags_map) {

api/api-doc/storage_service.json

@@ -3051,7 +3051,7 @@
                   },
                   {
                      "name":"incremental_mode",
-                     "description":"Set the incremental repair mode. Can be 'disabled', 'incremental', or 'full'. 'incremental': The incremental repair logic is enabled. Unrepaired sstables will be included for repair. Repaired sstables will be skipped. The incremental repair states will be updated after repair. 'full': The incremental repair logic is enabled. Both repaired and unrepaired sstables will be included for repair. The incremental repair states will be updated after repair. 'disabled': The incremental repair logic is disabled completely. The incremental repair states, e.g., repaired_at in sstables and sstables_repaired_at in the system.tablets table, will not be updated after repair. When the option is not provided, it defaults to 'disabled' mode.",
+                     "description":"Set the incremental repair mode. Can be 'disabled', 'incremental', or 'full'. 'incremental': The incremental repair logic is enabled. Unrepaired sstables will be included for repair. Repaired sstables will be skipped. The incremental repair states will be updated after repair. 'full': The incremental repair logic is enabled. Both repaired and unrepaired sstables will be included for repair. The incremental repair states will be updated after repair. 'disabled': The incremental repair logic is disabled completely. The incremental repair states, e.g., repaired_at in sstables and sstables_repaired_at in the system.tablets table, will not be updated after repair. When the option is not provided, it defaults to incremental mode.",
                      "required":false,
                      "allowMultiple":false,
                      "type":"string",

auth/allow_all_authenticator.cc

@@ -9,6 +9,7 @@
 #include "auth/allow_all_authenticator.hh"
 
 #include "service/migration_manager.hh"
+#include "utils/alien_worker.hh"
 #include "utils/class_registrator.hh"
 
 namespace auth {
@@ -22,6 +23,7 @@ static const class_registrator<
         cql3::query_processor&,
         ::service::raft_group0_client&,
         ::service::migration_manager&,
-        cache&> registration("org.apache.cassandra.auth.AllowAllAuthenticator");
+        cache&,
+        utils::alien_worker&> registration("org.apache.cassandra.auth.AllowAllAuthenticator");
 
 }

auth/allow_all_authenticator.hh

@@ -14,6 +14,7 @@
 #include "auth/authenticator.hh"
 #include "auth/cache.hh"
 #include "auth/common.hh"
+#include "utils/alien_worker.hh"
 
 namespace cql3 {
 class query_processor;
@@ -29,7 +30,7 @@ extern const std::string_view allow_all_authenticator_name;
 
 class allow_all_authenticator final : public authenticator {
 public:
-    allow_all_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&) {
+    allow_all_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&, utils::alien_worker&) {
     }
 
     virtual future<> start() override {

auth/certificate_authenticator.cc

@@ -8,7 +8,6 @@
  */
 
 #include "auth/certificate_authenticator.hh"
-#include "auth/cache.hh"
 
 #include <boost/regex.hpp>
 #include <fmt/ranges.h>
@@ -35,13 +34,13 @@ static const class_registrator<auth::authenticator
     , cql3::query_processor&
     , ::service::raft_group0_client&
     , ::service::migration_manager&
-    , auth::cache&> cert_auth_reg(CERT_AUTH_NAME);
+    , utils::alien_worker&> cert_auth_reg(CERT_AUTH_NAME);
 
 enum class auth::certificate_authenticator::query_source {
     subject, altname
 };
 
-auth::certificate_authenticator::certificate_authenticator(cql3::query_processor& qp, ::service::raft_group0_client&, ::service::migration_manager&, auth::cache&)
+auth::certificate_authenticator::certificate_authenticator(cql3::query_processor& qp, ::service::raft_group0_client&, ::service::migration_manager&, utils::alien_worker&)
     : _queries([&] {
         auto& conf = qp.db().get_config();
         auto queries = conf.auth_certificate_role_queries();
@@ -76,9 +75,9 @@ auth::certificate_authenticator::certificate_authenticator(cql3::query_processor
                         throw std::invalid_argument(fmt::format("Invalid source: {}", map.at(cfg_source_attr)));
                     }
                     continue;
-                } catch (const std::out_of_range&) {
+                } catch (std::out_of_range&) {
                     // just fallthrough
-                } catch (const boost::regex_error&) {
+                } catch (boost::regex_error&) {
                     std::throw_with_nested(std::invalid_argument(fmt::format("Invalid query expression: {}", map.at(cfg_query_attr))));
                 }
             }

auth/certificate_authenticator.hh

@@ -10,6 +10,7 @@
 #pragma once
 
 #include "auth/authenticator.hh"
+#include "utils/alien_worker.hh"
 #include <boost/regex_fwd.hpp>  // IWYU pragma: keep
 
 namespace cql3 {
@@ -25,15 +26,13 @@ class raft_group0_client;
 
 namespace auth {
 
-class cache;
-
 extern const std::string_view certificate_authenticator_name;
 
 class certificate_authenticator : public authenticator {
     enum class query_source;
     std::vector<std::pair<query_source, boost::regex>> _queries;
 public:
-    certificate_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&);
+    certificate_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, utils::alien_worker&);
     ~certificate_authenticator();
 
     future<> start() override;

auth/common.cc

@@ -94,7 +94,7 @@ static future<> create_legacy_metadata_table_if_missing_impl(
         try {
             co_return co_await mm.announce(co_await ::service::prepare_new_column_family_announcement(qp.proxy(), table, ts),
                     std::move(group0_guard), format("auth: create {} metadata table", table->cf_name()));
-        } catch (const exceptions::already_exists_exception&) {}
+        } catch (exceptions::already_exists_exception&) {}
     }
 }
 

auth/default_authorizer.cc

@@ -256,7 +256,7 @@ future<> default_authorizer::revoke_all(std::string_view role_name, ::service::g
         } else {
             co_await collect_mutations(_qp, mc, query, {sstring(role_name)});
         }
-    } catch (const exceptions::request_execution_exception& e) {
+    } catch (exceptions::request_execution_exception& e) {
         alogger.warn("CassandraAuthorizer failed to revoke all permissions of {}: {}", role_name, e);
     }
 }
@@ -293,13 +293,13 @@ future<> default_authorizer::revoke_all_legacy(const resource& resource) {
                                 [resource](auto ep) {
                     try {
                         std::rethrow_exception(ep);
-                    } catch (const exceptions::request_execution_exception& e) {
+                    } catch (exceptions::request_execution_exception& e) {
                         alogger.warn("CassandraAuthorizer failed to revoke all permissions on {}: {}", resource, e);
                     }
 
                 });
             });
-        } catch (const exceptions::request_execution_exception& e) {
+        } catch (exceptions::request_execution_exception& e) {
             alogger.warn("CassandraAuthorizer failed to revoke all permissions on {}: {}", resource, e);
             return make_ready_future();
         }

auth/password_authenticator.cc

@@ -49,7 +49,8 @@ static const class_registrator<
         cql3::query_processor&,
         ::service::raft_group0_client&,
         ::service::migration_manager&,
-        cache&> password_auth_reg("org.apache.cassandra.auth.PasswordAuthenticator");
+        cache&,
+        utils::alien_worker&> password_auth_reg("org.apache.cassandra.auth.PasswordAuthenticator");
 
 static thread_local auto rng_for_salt = std::default_random_engine(std::random_device{}());
 
@@ -63,13 +64,14 @@ std::string password_authenticator::default_superuser(const db::config& cfg) {
 password_authenticator::~password_authenticator() {
 }
 
-password_authenticator::password_authenticator(cql3::query_processor& qp, ::service::raft_group0_client& g0, ::service::migration_manager& mm, cache& cache)
+password_authenticator::password_authenticator(cql3::query_processor& qp, ::service::raft_group0_client& g0, ::service::migration_manager& mm, cache& cache, utils::alien_worker& hashing_worker)
     : _qp(qp)
     , _group0_client(g0)
     , _migration_manager(mm)
     , _cache(cache)
     , _stopped(make_ready_future<>()) 
     , _superuser(default_superuser(qp.db().get_config()))
+    , _hashing_worker(hashing_worker)
 {}
 
 static bool has_salted_hash(const cql3::untyped_result_set_row& row) {
@@ -328,18 +330,20 @@ future<authenticated_user> password_authenticator::authenticate(
             }
             salted_hash = role->salted_hash;
         }
-        const bool password_match = co_await passwords::check(password, *salted_hash);
+        const bool password_match = co_await _hashing_worker.submit<bool>([password = std::move(password), salted_hash] {
+            return passwords::check(password, *salted_hash);
+        });
         if (!password_match) {
             throw exceptions::authentication_exception("Username and/or password are incorrect");
         }
         co_return username;
-    } catch (const std::system_error &) {
+    } catch (std::system_error &) {
         std::throw_with_nested(exceptions::authentication_exception("Could not verify password"));
-    } catch (const exceptions::request_execution_exception& e) {
+    } catch (exceptions::request_execution_exception& e) {
         std::throw_with_nested(exceptions::authentication_exception(e.what()));
-    } catch (const exceptions::authentication_exception& e) {
+    } catch (exceptions::authentication_exception& e) {
         std::throw_with_nested(e);
-    } catch (const exceptions::unavailable_exception& e) {
+    } catch (exceptions::unavailable_exception& e) {
         std::throw_with_nested(exceptions::authentication_exception(e.get_message()));
     } catch (...) {
         std::throw_with_nested(exceptions::authentication_exception("authentication failed"));

auth/password_authenticator.hh

@@ -18,6 +18,7 @@
 #include "auth/passwords.hh"
 #include "auth/cache.hh"
 #include "service/raft/raft_group0_client.hh"
+#include "utils/alien_worker.hh"
 
 namespace db {
     class config;
@@ -48,12 +49,13 @@ class password_authenticator : public authenticator {
     shared_promise<> _superuser_created_promise;
     // We used to also support bcrypt, SHA-256, and MD5 (ref. scylladb#24524).
     constexpr static auth::passwords::scheme _scheme = passwords::scheme::sha_512;
+    utils::alien_worker& _hashing_worker;
 
 public:
     static db::consistency_level consistency_for_user(std::string_view role_name);
     static std::string default_superuser(const db::config&);
 
-    password_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&);
+    password_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&, utils::alien_worker&);
 
     ~password_authenticator();
 

auth/passwords.cc

@@ -7,8 +7,6 @@
  */
 
 #include "auth/passwords.hh"
-#include "utils/crypt_sha512.hh"
-#include <seastar/core/coroutine.hh>
 
 #include <cerrno>
 
@@ -23,46 +21,25 @@ static thread_local crypt_data tlcrypt = {};
 
 namespace detail {
 
-void verify_hashing_output(const char * res) {
-    if (!res || (res[0] == '*')) {
-        throw std::system_error(errno, std::system_category());
-    }
-}
-
 void verify_scheme(scheme scheme) {
     const sstring random_part_of_salt = "aaaabbbbccccdddd";
 
     const sstring salt = sstring(prefix_for_scheme(scheme)) + random_part_of_salt;
     const char* e = crypt_r("fisk", salt.c_str(), &tlcrypt);
-    try {
-        verify_hashing_output(e);
-    } catch (const std::system_error& ex) {
-        throw no_supported_schemes();
+
+    if (e && (e[0] != '*')) {
+        return;
     }
+
+    throw no_supported_schemes();
 }
 
 sstring hash_with_salt(const sstring& pass, const sstring& salt) {
     auto res = crypt_r(pass.c_str(), salt.c_str(), &tlcrypt);
-    verify_hashing_output(res);
-    return res;
-}
-
-seastar::future<sstring> hash_with_salt_async(const sstring& pass, const sstring& salt) {
-    sstring res;
-    // Only SHA-512 hashes for passphrases shorter than 256 bytes can be computed using
-    // the __crypt_sha512 method. For other computations, we fall back to the
-    // crypt_r implementation from `<crypt.h>`, which can stall.
-    if (salt.starts_with(prefix_for_scheme(scheme::sha_512)) && pass.size() <= 255) {
-        char buf[128];
-        const char * output_ptr = co_await __crypt_sha512(pass.c_str(), salt.c_str(), buf);
-        verify_hashing_output(output_ptr);
-        res = output_ptr;
-    } else {
-        const char * output_ptr = crypt_r(pass.c_str(), salt.c_str(), &tlcrypt);
-        verify_hashing_output(output_ptr);
-        res = output_ptr;
+    if (!res || (res[0] == '*')) {
+        throw std::system_error(errno, std::system_category());
     }
-    co_return res;
+    return res;
 }
 
 std::string_view prefix_for_scheme(scheme c) noexcept {
@@ -81,9 +58,8 @@ no_supported_schemes::no_supported_schemes()
         : std::runtime_error("No allowed hashing schemes are supported on this system") {
 }
 
-seastar::future<bool> check(const sstring& pass, const sstring& salted_hash) {
-    const auto pwd_hash = co_await detail::hash_with_salt_async(pass, salted_hash);
-    co_return pwd_hash == salted_hash;
+bool check(const sstring& pass, const sstring& salted_hash) {
+    return detail::hash_with_salt(pass, salted_hash) == salted_hash;
 }
 
 } // namespace auth::passwords

auth/passwords.hh

@@ -11,7 +11,6 @@
 #include <random>
 #include <stdexcept>
 
-#include <seastar/core/future.hh>
 #include <seastar/core/sstring.hh>
 
 #include "seastarx.hh"
@@ -76,19 +75,10 @@ sstring generate_salt(RandomNumberEngine& g, scheme scheme) {
 
 ///
 /// Hash a password combined with an implementation-specific salt string.
-/// Deprecated in favor of `hash_with_salt_async`.
 ///
 /// \throws \ref std::system_error when an unexpected implementation-specific error occurs.
 ///
-[[deprecated("Use hash_with_salt_async instead")]] sstring hash_with_salt(const sstring& pass, const sstring& salt);
-
-///
-/// Async version of `hash_with_salt` that returns a future.
-/// If possible, hashing uses `coroutine::maybe_yield` to prevent reactor stalls.
-///
-/// \throws \ref std::system_error when an unexpected implementation-specific error occurs.
-///
-seastar::future<sstring> hash_with_salt_async(const sstring& pass, const sstring& salt);
+sstring hash_with_salt(const sstring& pass, const sstring& salt);
 
 } // namespace detail
 
@@ -117,6 +107,6 @@ sstring hash(const sstring& pass, RandomNumberEngine& g, scheme scheme) {
 ///
 /// \throws \ref std::system_error when an unexpected implementation-specific error occurs.
 ///
-seastar::future<bool> check(const sstring& pass, const sstring& salted_hash);
+bool check(const sstring& pass, const sstring& salted_hash);
 
 } // namespace auth::passwords

auth/saslauthd_authenticator.cc

@@ -35,9 +35,10 @@ static const class_registrator<
         cql3::query_processor&,
         ::service::raft_group0_client&,
         ::service::migration_manager&,
-        cache&> saslauthd_auth_reg("com.scylladb.auth.SaslauthdAuthenticator");
+        cache&,
+        utils::alien_worker&> saslauthd_auth_reg("com.scylladb.auth.SaslauthdAuthenticator");
 
-saslauthd_authenticator::saslauthd_authenticator(cql3::query_processor& qp, ::service::raft_group0_client&, ::service::migration_manager&, cache&)
+saslauthd_authenticator::saslauthd_authenticator(cql3::query_processor& qp, ::service::raft_group0_client&, ::service::migration_manager&, cache&, utils::alien_worker&)
     : _socket_path(qp.db().get_config().saslauthd_socket_path())
 {}
 

auth/saslauthd_authenticator.hh

@@ -12,6 +12,7 @@
 
 #include "auth/authenticator.hh"
 #include "auth/cache.hh"
+#include "utils/alien_worker.hh"
 
 namespace cql3 {
 class query_processor;
@@ -29,7 +30,7 @@ namespace auth {
 class saslauthd_authenticator : public authenticator {
     sstring _socket_path; ///< Path to the domain socket on which saslauthd is listening.
 public:
-    saslauthd_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&);
+    saslauthd_authenticator(cql3::query_processor&, ::service::raft_group0_client&, ::service::migration_manager&, cache&,utils::alien_worker&);
 
     future<> start() override;
 

auth/service.cc

@@ -191,7 +191,8 @@ service::service(
         ::service::migration_manager& mm,
         const service_config& sc,
         maintenance_socket_enabled used_by_maintenance_socket,
-        cache& cache)
+        cache& cache,
+        utils::alien_worker& hashing_worker)
             : service(
                       std::move(c),
                       cache,
@@ -199,7 +200,7 @@ service::service(
                       g0,
                       mn,
                       create_object<authorizer>(sc.authorizer_java_name, qp, g0, mm),
-                      create_object<authenticator>(sc.authenticator_java_name, qp, g0, mm, cache),
+                      create_object<authenticator>(sc.authenticator_java_name, qp, g0, mm, cache, hashing_worker),
                       create_object<role_manager>(sc.role_manager_java_name, qp, g0, mm, cache),
                       used_by_maintenance_socket) {
 }
@@ -225,7 +226,7 @@ future<> service::create_legacy_keyspace_if_missing(::service::migration_manager
             try {
                 co_return co_await mm.announce(::service::prepare_new_keyspace_announcement(db.real_database(), ksm, ts),
                         std::move(group0_guard), seastar::format("auth_service: create {} keyspace", meta::legacy::AUTH_KS));
-            } catch (const ::service::group0_concurrent_modification&) {
+            } catch (::service::group0_concurrent_modification&) {
                 log.info("Concurrent operation is detected while creating {} keyspace, retrying.", meta::legacy::AUTH_KS);
             }
         }

auth/service.hh

@@ -27,6 +27,7 @@
 #include "cql3/description.hh"
 #include "seastarx.hh"
 #include "service/raft/raft_group0_client.hh"
+#include "utils/alien_worker.hh"
 #include "utils/observable.hh"
 #include "utils/serialized_action.hh"
 #include "service/maintenance_mode.hh"
@@ -130,7 +131,8 @@ public:
             ::service::migration_manager&,
             const service_config&,
             maintenance_socket_enabled,
-            cache&);
+            cache&,
+            utils::alien_worker&);
 
     future<> start(::service::migration_manager&, db::system_keyspace&);
 

auth/standard_role_manager.cc

@@ -192,7 +192,7 @@ future<> standard_role_manager::legacy_create_default_role_if_missing() {
                 {_superuser},
                 cql3::query_processor::cache_internal::no).discard_result();
         log.info("Created default superuser role '{}'.", _superuser);
-    } catch (const exceptions::unavailable_exception& e) {
+    } catch(const exceptions::unavailable_exception& e) {
         log.warn("Skipped default role setup: some nodes were not ready; will retry");
         throw e;
     }

auth/transitional.cc

@@ -38,8 +38,8 @@ class transitional_authenticator : public authenticator {
 public:
     static const sstring PASSWORD_AUTHENTICATOR_NAME;
 
-    transitional_authenticator(cql3::query_processor& qp, ::service::raft_group0_client& g0, ::service::migration_manager& mm, cache& cache)
-            : transitional_authenticator(std::make_unique<password_authenticator>(qp, g0, mm, cache)) {
+    transitional_authenticator(cql3::query_processor& qp, ::service::raft_group0_client& g0, ::service::migration_manager& mm, cache& cache, utils::alien_worker& hashing_worker)
+            : transitional_authenticator(std::make_unique<password_authenticator>(qp, g0, mm, cache, hashing_worker)) {
     }
     transitional_authenticator(std::unique_ptr<authenticator> a)
             : _authenticator(std::move(a)) {
@@ -81,7 +81,7 @@ public:
         }).handle_exception([](auto ep) {
             try {
                 std::rethrow_exception(ep);
-            } catch (const exceptions::authentication_exception&) {
+            } catch (exceptions::authentication_exception&) {
                 // return anon user
                 return make_ready_future<authenticated_user>(anonymous_user());
             }
@@ -126,7 +126,7 @@ public:
             virtual bytes evaluate_response(bytes_view client_response) override {
                 try {
                     return _sasl->evaluate_response(client_response);
-                } catch (const exceptions::authentication_exception&) {
+                } catch (exceptions::authentication_exception&) {
                     _complete = true;
                     return {};
                 }
@@ -141,7 +141,7 @@ public:
                     return _sasl->get_authenticated_user().handle_exception([](auto ep) {
                         try {
                             std::rethrow_exception(ep);
-                        } catch (const exceptions::authentication_exception&) {
+                        } catch (exceptions::authentication_exception&) {
                             // return anon user
                             return make_ready_future<authenticated_user>(anonymous_user());
                         }
@@ -241,7 +241,8 @@ static const class_registrator<
         cql3::query_processor&,
         ::service::raft_group0_client&,
         ::service::migration_manager&,
-        auth::cache&> transitional_authenticator_reg(auth::PACKAGE_NAME + "TransitionalAuthenticator");
+        auth::cache&,
+        utils::alien_worker&> transitional_authenticator_reg(auth::PACKAGE_NAME + "TransitionalAuthenticator");
 
 static const class_registrator<
         auth::authorizer,

configure.py

@@ -859,7 +859,6 @@ scylla_core = (['message/messaging_service.cc',
                 'utils/alien_worker.cc',
                 'utils/array-search.cc',
                 'utils/base64.cc',
-                'utils/crypt_sha512.cc',
                 'utils/logalloc.cc',
                 'utils/large_bitset.cc',
                 'utils/buffer_input_stream.cc',
@@ -1063,6 +1062,7 @@ scylla_core = (['message/messaging_service.cc',
                 'db/hints/resource_manager.cc',
                 'db/hints/sync_point.cc',
                 'db/large_data_handler.cc',
+                'db/legacy_schema_migrator.cc',
                 'db/marshal/type_parser.cc',
                 'db/per_partition_rate_limit_options.cc',
                 'db/rate_limiter.cc',
@@ -1480,6 +1480,7 @@ deps = {
 
 pure_boost_tests = set([
     'test/boost/anchorless_list_test',
+    'test/boost/auth_passwords_test',
     'test/boost/auth_resource_test',
     'test/boost/big_decimal_test',
     'test/boost/caching_options_test',

cql3/statements/batch_statement.cc

@@ -331,7 +331,7 @@ future<shared_ptr<cql_transport::messages::result_message>> batch_statement::exe
     if (!cl_for_paxos) [[unlikely]] {
         return make_exception_future<shared_ptr<cql_transport::messages::result_message>>(std::move(cl_for_paxos).assume_error());
     }
-    std::unique_ptr<cas_request> request;
+    seastar::shared_ptr<cas_request> request;
     schema_ptr schema;
 
     db::timeout_clock::time_point now = db::timeout_clock::now();
@@ -354,9 +354,9 @@ future<shared_ptr<cql_transport::messages::result_message>> batch_statement::exe
         if (keys.empty()) {
             continue;
         }
-        if (!request) {
+        if (request.get() == nullptr) {
             schema = statement.s;
-            request = std::make_unique<cas_request>(schema, std::move(keys));
+            request = seastar::make_shared<cas_request>(schema, std::move(keys));
         } else if (keys.size() != 1 || keys.front().equal(request->key().front(), dht::ring_position_comparator(*schema)) == false) {
             throw exceptions::invalid_request_exception("BATCH with conditions cannot span multiple partitions");
         }
@@ -366,7 +366,7 @@ future<shared_ptr<cql_transport::messages::result_message>> batch_statement::exe
 
         request->add_row_update(statement, std::move(ranges), std::move(json_cache), statement_options);
     }
-    if (!request) {
+    if (request.get() == nullptr) {
         throw exceptions::invalid_request_exception(format("Unrestricted partition key in a conditional BATCH"));
     }
 
@@ -377,10 +377,9 @@ future<shared_ptr<cql_transport::messages::result_message>> batch_statement::exe
             );
     }
 
-    auto* request_ptr = request.get();
-    return qp.proxy().cas(schema, std::move(cas_shard), *request_ptr, request->read_command(qp), request->key(),
+    return qp.proxy().cas(schema, std::move(cas_shard), request, request->read_command(qp), request->key(),
             {read_timeout, qs.get_permit(), qs.get_client_state(), qs.get_trace_state()},
-            std::move(cl_for_paxos).assume_value(), cl_for_learn, batch_timeout, cas_timeout).then([this, request = std::move(request)] (bool is_applied) {
+            std::move(cl_for_paxos).assume_value(), cl_for_learn, batch_timeout, cas_timeout).then([this, request] (bool is_applied) {
         return request->build_cas_result_set(_metadata, _columns_of_cas_result_set, is_applied);
     });
 }

cql3/statements/modification_statement.cc

@@ -401,8 +401,7 @@ modification_statement::execute_with_condition(query_processor& qp, service::que
                     type.is_update() ? "update" : "deletion"));
     }
 
-    auto request = std::make_unique<cas_request>(s, std::move(keys));
-    auto* request_ptr = request.get();
+    auto request = seastar::make_shared<cas_request>(s, std::move(keys));
     // cas_request can be used for batches as well single statements; Here we have just a single
     // modification in the list of CAS commands, since we're handling single-statement execution.
     request->add_row_update(*this, std::move(ranges), std::move(json_cache), options);
@@ -428,9 +427,9 @@ modification_statement::execute_with_condition(query_processor& qp, service::que
         tablet_info = erm->check_locality(token);
     }
 
-    return qp.proxy().cas(s, std::move(cas_shard), *request_ptr, request->read_command(qp), request->key(),
+    return qp.proxy().cas(s, std::move(cas_shard), request, request->read_command(qp), request->key(),
             {read_timeout, qs.get_permit(), qs.get_client_state(), qs.get_trace_state()},
-            std::move(cl_for_paxos).assume_value(), cl_for_learn, statement_timeout, cas_timeout).then([this, request = std::move(request), tablet_replicas = std::move(tablet_info->tablet_replicas), token_range = tablet_info->token_range] (bool is_applied) {
+            std::move(cl_for_paxos).assume_value(), cl_for_learn, statement_timeout, cas_timeout).then([this, request, tablet_replicas = std::move(tablet_info->tablet_replicas), token_range = tablet_info->token_range] (bool is_applied) {
         auto result = request->build_cas_result_set(_metadata, _columns_of_cas_result_set, is_applied);
         result->add_tablet_info(tablet_replicas, token_range);
         return result;

db/CMakeLists.txt

@@ -10,6 +10,7 @@ target_sources(db
     schema_applier.cc
     schema_tables.cc
     cql_type_parser.cc
+    legacy_schema_migrator.cc
     commitlog/commitlog.cc
     commitlog/commitlog_replayer.cc
     commitlog/commitlog_entry.cc

db/commitlog/commitlog_replayer.cc

@@ -165,7 +165,7 @@ future<> db::commitlog_replayer::impl::init() {
 
 future<db::commitlog_replayer::impl::stats>
 db::commitlog_replayer::impl::recover(const commitlog::descriptor& d, const commitlog::replay_state& rpstate) const {
-    SCYLLA_ASSERT(_column_mappings.local_is_initialized());
+    scylla_assert(_column_mappings.local_is_initialized());
 
     replay_position rp{d};
     auto gp = min_pos(rp.shard_id());

db/corrupt_data_handler.cc

@@ -10,6 +10,7 @@
 #include "reader_concurrency_semaphore.hh"
 #include "replica/database.hh"
 #include "utils/UUID_gen.hh"
+#include "utils/assert.hh"
 
 static logging::logger corrupt_data_logger("corrupt_data");
 
@@ -75,14 +76,14 @@ future<corrupt_data_handler::entry_id> system_table_corrupt_data_handler::do_rec
 
     auto set_cell_raw = [this, &entry_row, &corrupt_data_schema, timestamp] (const char* cell_name, managed_bytes cell_value) {
         auto cdef = corrupt_data_schema->get_column_definition(cell_name);
-        SCYLLA_ASSERT(cdef);
+        scylla_assert(cdef);
 
         entry_row.cells().apply(*cdef, atomic_cell::make_live(*cdef->type, timestamp, cell_value, _entry_ttl));
     }; 
 
     auto set_cell = [this, &entry_row, &corrupt_data_schema, timestamp] (const char* cell_name, data_value cell_value) {
         auto cdef = corrupt_data_schema->get_column_definition(cell_name);
-        SCYLLA_ASSERT(cdef);
+        scylla_assert(cdef);
 
         entry_row.cells().apply(*cdef, atomic_cell::make_live(*cdef->type, timestamp, cell_value.serialize_nonnull(), _entry_ttl));
     };

db/large_data_handler.cc

@@ -39,7 +39,7 @@ large_data_handler::large_data_handler(uint64_t partition_threshold_bytes, uint6
 }
 
 future<large_data_handler::partition_above_threshold> large_data_handler::maybe_record_large_partitions(const sstables::sstable& sst, const sstables::key& key, uint64_t partition_size, uint64_t rows, uint64_t range_tombstones, uint64_t dead_rows) {
-    SCYLLA_ASSERT(running());
+    scylla_assert(running());
     partition_above_threshold above_threshold{partition_size > _partition_threshold_bytes, rows > _rows_count_threshold};
     static_assert(std::is_same_v<decltype(above_threshold.size), bool>);
     _stats.partitions_bigger_than_threshold += above_threshold.size; // increment if true
@@ -83,7 +83,7 @@ sstring large_data_handler::sst_filename(const sstables::sstable& sst) {
 }
 
 future<> large_data_handler::maybe_delete_large_data_entries(sstables::shared_sstable sst) {
-    SCYLLA_ASSERT(running());
+    scylla_assert(running());
     auto schema = sst->get_schema();
     auto filename = sst_filename(*sst);
     using ldt = sstables::large_data_type;
@@ -247,7 +247,7 @@ future<> cql_table_large_data_handler::record_large_rows(const sstables::sstable
 
 future<> cql_table_large_data_handler::delete_large_data_entries(const schema& s, sstring sstable_name, std::string_view large_table_name) const {
     auto sys_ks = _sys_ks.get_permit();
-    SCYLLA_ASSERT(sys_ks);
+    scylla_assert(sys_ks);
     const sstring req =
             seastar::format("DELETE FROM system.{} WHERE keyspace_name = ? AND table_name = ? AND sstable_name = ?",
                     large_table_name);

db/large_data_handler.hh

@@ -80,7 +80,7 @@ public:
 
     future<bool> maybe_record_large_rows(const sstables::sstable& sst, const sstables::key& partition_key,
             const clustering_key_prefix* clustering_key, uint64_t row_size) {
-        SCYLLA_ASSERT(running());
+        scylla_assert(running());
         if (row_size > _row_threshold_bytes) [[unlikely]] {
             return with_sem([&sst, &partition_key, clustering_key, row_size, this] {
                 return record_large_rows(sst, partition_key, clustering_key, row_size);
@@ -100,7 +100,7 @@ public:
 
     future<bool> maybe_record_large_cells(const sstables::sstable& sst, const sstables::key& partition_key,
             const clustering_key_prefix* clustering_key, const column_definition& cdef, uint64_t cell_size, uint64_t collection_elements) {
-        SCYLLA_ASSERT(running());
+        scylla_assert(running());
         if (cell_size > _cell_threshold_bytes || collection_elements > _collection_elements_count_threshold) [[unlikely]] {
             return with_sem([&sst, &partition_key, clustering_key, &cdef, cell_size, collection_elements, this] {
                 return record_large_cells(sst, partition_key, clustering_key, cdef, cell_size, collection_elements);

db/legacy_schema_migrator.cc

@@ -0,0 +1,602 @@
+/*
+ * Modified by ScyllaDB
+ * Copyright (C) 2017-present ScyllaDB
+ */
+
+/*
+ * SPDX-License-Identifier: (LicenseRef-ScyllaDB-Source-Available-1.0 and Apache-2.0)
+ */
+
+// Since Scylla 2.0, we use system tables whose schemas were introduced in
+// Cassandra 3. If Scylla boots to find a data directory with system tables
+// with older schemas - produced by pre-2.0 Scylla or by pre-3.0 Cassandra,
+// we need to migrate these old tables to the new format.
+//
+// We provide here a function, db::legacy_schema_migrator::migrate(),
+// for a one-time migration from old to new system tables. The function
+// reads old system tables, write them back in the new format, and finally
+// delete the old system tables. Scylla's main should call this function and
+// wait for the returned future, before starting to serve the database.
+
+#include <boost/iterator/filter_iterator.hpp>
+#include <seastar/core/future-util.hh>
+#include <seastar/util/log.hh>
+#include <map>
+#include <unordered_set>
+#include <chrono>
+
+#include "replica/database.hh"
+#include "legacy_schema_migrator.hh"
+#include "system_keyspace.hh"
+#include "schema_tables.hh"
+#include "schema/schema_builder.hh"
+#include "service/storage_proxy.hh"
+#include "utils/rjson.hh"
+#include "cql3/query_processor.hh"
+#include "cql3/untyped_result_set.hh"
+#include "cql3/util.hh"
+#include "cql3/statements/property_definitions.hh"
+
+static seastar::logger mlogger("legacy_schema_migrator");
+
+namespace db {
+namespace legacy_schema_migrator {
+
+// local data carriers
+
+class migrator {
+public:
+    static const std::unordered_set<sstring> legacy_schema_tables;
+
+    migrator(sharded<service::storage_proxy>& sp, sharded<replica::database>& db, sharded<db::system_keyspace>& sys_ks, cql3::query_processor& qp)
+                    : _sp(sp), _db(db), _sys_ks(sys_ks), _qp(qp) {
+    }
+    migrator(migrator&&) = default;
+
+    typedef db_clock::time_point time_point;
+
+    // TODO: we don't support triggers.
+    // this is a placeholder.
+    struct trigger {
+        time_point timestamp;
+        sstring name;
+        std::unordered_map<sstring, sstring> options;
+    };
+
+    struct table {
+        time_point timestamp;
+        schema_ptr metadata;
+        std::vector<trigger> triggers;
+    };
+
+    struct type {
+        time_point timestamp;
+        user_type metadata;
+    };
+
+    struct function {
+        time_point timestamp;
+        sstring ks_name;
+        sstring fn_name;
+        std::vector<sstring> arg_names;
+        std::vector<sstring> arg_types;
+        sstring return_type;
+        bool called_on_null_input;
+        sstring language;
+        sstring body;
+    };
+
+    struct aggregate {
+        time_point timestamp;
+        sstring ks_name;
+        sstring fn_name;
+        std::vector<sstring> arg_names;
+        std::vector<sstring> arg_types;
+        sstring return_type;
+        sstring final_func;
+        sstring initcond;
+        sstring state_func;
+        sstring state_type;
+    };
+
+    struct keyspace {
+        time_point timestamp;
+        sstring name;
+        bool durable_writes;
+        std::map<sstring, sstring> replication_params;
+
+        std::vector<table> tables;
+        std::vector<type> types;
+        std::vector<function> functions;
+        std::vector<aggregate> aggregates;
+    };
+
+    class unsupported_feature : public std::runtime_error {
+    public:
+        using runtime_error::runtime_error;
+    };
+
+    static sstring fmt_query(const char* fmt, const char* table) {
+        return fmt::format(fmt::runtime(fmt), db::system_keyspace::NAME, table);
+    }
+
+    typedef ::shared_ptr<cql3::untyped_result_set> result_set_type;
+    typedef const cql3::untyped_result_set::row row_type;
+
+    future<> read_table(keyspace& dst, sstring cf_name, time_point timestamp) {
+        auto fmt = "SELECT * FROM {}.{} WHERE keyspace_name = ? AND columnfamily_name = ?";
+        auto tq = fmt_query(fmt, db::system_keyspace::legacy::COLUMNFAMILIES);
+        auto cq = fmt_query(fmt, db::system_keyspace::legacy::COLUMNS);
+        auto zq = fmt_query(fmt, db::system_keyspace::legacy::TRIGGERS);
+
+        typedef std::tuple<future<result_set_type>, future<result_set_type>, future<result_set_type>, future<db::schema_tables::legacy::schema_mutations>> result_tuple;
+
+        return when_all(_qp.execute_internal(tq, { dst.name, cf_name }, cql3::query_processor::cache_internal::yes),
+                        _qp.execute_internal(cq, { dst.name, cf_name }, cql3::query_processor::cache_internal::yes),
+                        _qp.execute_internal(zq, { dst.name, cf_name }, cql3::query_processor::cache_internal::yes),
+                        db::schema_tables::legacy::read_table_mutations(_sp, dst.name, cf_name, db::system_keyspace::legacy::column_families()))
+                    .then([&dst, cf_name, timestamp](result_tuple&& t) {
+
+            result_set_type tables = std::get<0>(t).get();
+            result_set_type columns = std::get<1>(t).get();
+            result_set_type triggers = std::get<2>(t).get();
+            db::schema_tables::legacy::schema_mutations sm = std::get<3>(t).get();
+
+            row_type& td = tables->one();
+
+            auto ks_name = td.get_as<sstring>("keyspace_name");
+            auto cf_name = td.get_as<sstring>("columnfamily_name");
+            auto id = table_id(td.get_or("cf_id", generate_legacy_id(ks_name, cf_name).uuid()));
+
+            schema_builder builder(dst.name, cf_name, id);
+
+            builder.with_version(sm.digest());
+
+            cf_type cf = sstring_to_cf_type(td.get_or("type", sstring("standard")));
+            if (cf == cf_type::super) {
+                fail(unimplemented::cause::SUPER);
+            }
+
+            auto comparator = td.get_as<sstring>("comparator");
+            bool is_compound = cell_comparator::check_compound(comparator);
+            builder.set_is_compound(is_compound);
+            cell_comparator::read_collections(builder, comparator);
+
+            bool filter_sparse = false;
+
+            data_type default_validator = {};
+            if (td.has("default_validator")) {
+                default_validator = db::schema_tables::parse_type(td.get_as<sstring>("default_validator"));
+                if (default_validator->is_counter()) {
+                    builder.set_is_counter(true);
+                }
+                builder.set_default_validation_class(default_validator);
+            }
+
+            /*
+             * Determine whether or not the table is *really* dense
+             * We cannot trust is_dense value of true (see CASSANDRA-11502, that fixed the issue for 2.2 only, and not retroactively),
+             * but we can trust is_dense value of false.
+             */
+            auto is_dense = td.get_opt<bool>("is_dense");
+            if (!is_dense || *is_dense) {
+                is_dense = [&] {
+                    /*
+                     * As said above, this method is only here because we need to deal with thrift upgrades.
+                     * Once a CF has been "upgraded", i.e. we've rebuilt and save its CQL3 metadata at least once,
+                     * then we'll have saved the "is_dense" value and will be good to go.
+                     *
+                     * But non-upgraded thrift CF (and pre-7744 CF) will have no value for "is_dense", so we need
+                     * to infer that information without relying on it in that case. And for the most part this is
+                     * easy, a CF that has at least one REGULAR definition is not dense. But the subtlety is that not
+                     * having a REGULAR definition may not mean dense because of CQL3 definitions that have only the
+                     * PRIMARY KEY defined.
+                     *
+                     * So we need to recognize those special case CQL3 table with only a primary key. If we have some
+                     * clustering columns, we're fine as said above. So the only problem is that we cannot decide for
+                     * sure if a CF without REGULAR columns nor CLUSTERING_COLUMN definition is meant to be dense, or if it
+                     * has been created in CQL3 by say:
+                     *    CREATE TABLE test (k int PRIMARY KEY)
+                     * in which case it should not be dense. However, we can limit our margin of error by assuming we are
+                     * in the latter case only if the comparator is exactly CompositeType(UTF8Type).
+                     */
+                    std::optional<column_id> max_cl_idx;
+                    const cql3::untyped_result_set::row * regular = nullptr;
+                    for (auto& row : *columns) {
+                        auto kind_str = row.get_as<sstring>("type");
+                        if (kind_str == "compact_value") {
+                            continue;
+                        }
+
+                        auto kind = db::schema_tables::deserialize_kind(kind_str);
+
+                        if (kind == column_kind::regular_column) {
+                            if (regular != nullptr) {
+                                return false;
+                            }
+                            regular = &row;
+                            continue;
+                        }
+                        if (kind == column_kind::clustering_key) {
+                            max_cl_idx = std::max(column_id(row.get_or("component_index", 0)), max_cl_idx.value_or(column_id()));
+                        }
+                    }
+
+                    auto is_cql3_only_pk_comparator = [](const sstring& comparator) {
+                        if (!cell_comparator::check_compound(comparator)) {
+                            return false;
+                        }
+                        // CMH. We don't have composites, nor a parser for it. This is a simple way of c
+                        // checking the same.
+                        auto comma = comparator.find(',');
+                        if (comma != sstring::npos) {
+                            return false;
+                        }
+                        auto off = comparator.find('(');
+                        auto end = comparator.find(')');
+
+                        return comparator.compare(off, end - off, utf8_type->name()) == 0;
+                    };
+
+                    if (max_cl_idx) {
+                        auto n = std::count(comparator.begin(), comparator.end(), ','); // num comp - 1
+                        return *max_cl_idx == n;
+                    }
+
+                    if (regular) {
+                        return false;
+                    }
+
+                    return !is_cql3_only_pk_comparator(comparator);
+
+                }();
+
+                // now, if switched to sparse, remove redundant compact_value column and the last clustering column,
+                // directly copying CASSANDRA-11502 logic. See CASSANDRA-11315.
+
+                filter_sparse = !*is_dense;
+            }
+            builder.set_is_dense(*is_dense);
+
+            auto is_cql = !*is_dense && is_compound;
+            auto is_static_compact = !*is_dense && !is_compound;
+
+            // org.apache.cassandra.schema.LegacySchemaMigrator#isEmptyCompactValueColumn
+            auto is_empty_compact_value = [](const cql3::untyped_result_set::row& column_row) {
+                auto kind_str = column_row.get_as<sstring>("type");
+                // Cassandra only checks for "compact_value", but Scylla generates "regular" instead (#2586)
+                return (kind_str == "compact_value" || kind_str == "regular")
+                       && column_row.get_as<sstring>("column_name").empty();
+            };
+
+            for (auto& row : *columns) {
+                auto kind_str = row.get_as<sstring>("type");
+                auto kind = db::schema_tables::deserialize_kind(kind_str);
+                auto component_index = kind > column_kind::clustering_key ? 0 : column_id(row.get_or("component_index", 0));
+                auto name = row.get_or<sstring>("column_name", sstring());
+                auto validator = db::schema_tables::parse_type(row.get_as<sstring>("validator"));
+
+                if (is_empty_compact_value(row)) {
+                    continue;
+                }
+
+                if (filter_sparse) {
+                    if (kind_str == "compact_value") {
+                        continue;
+                    }
+                    if (kind == column_kind::clustering_key) {
+                        if (cf == cf_type::super && component_index != 0) {
+                            continue;
+                        }
+                        if (cf != cf_type::super && !is_compound) {
+                            continue;
+                        }
+                    }
+                }
+
+                std::optional<index_metadata_kind> index_kind;
+                sstring index_name;
+                index_options_map options;
+                if (row.has("index_type")) {
+                    index_kind = schema_tables::deserialize_index_kind(row.get_as<sstring>("index_type"));
+                }
+                if (row.has("index_name")) {
+                    index_name = row.get_as<sstring>("index_name");
+                }
+                if (row.has("index_options")) {
+                    sstring index_options_str = row.get_as<sstring>("index_options");
+                    options = rjson::parse_to_map<index_options_map>(std::string_view(index_options_str));
+                    sstring type;
+                    auto i = options.find("index_keys");
+                    if (i != options.end()) {
+                        options.erase(i);
+                        type = "KEYS";
+                    }
+                    i = options.find("index_keys_and_values");
+                    if (i != options.end()) {
+                        options.erase(i);
+                        type = "KEYS_AND_VALUES";
+                    }
+                    if (type.empty()) {
+                        if (validator->is_collection() && validator->is_multi_cell()) {
+                            type = "FULL";
+                        } else {
+                            type = "VALUES";
+                        }
+                    }
+                    auto column = cql3::util::maybe_quote(name);
+                    options["target"] = validator->is_collection()
+                                    ? type + "(" + column + ")"
+                                    : column;
+                }
+                if (index_kind) {
+                    // Origin assumes index_name is always set, so let's do the same
+                    builder.with_index(index_metadata(index_name, options, *index_kind, index_metadata::is_local_index::no));
+                }
+
+                data_type column_name_type = [&] {
+                    if (is_static_compact && kind == column_kind::regular_column) {
+                        return db::schema_tables::parse_type(comparator);
+                    }
+                    return utf8_type;
+                }();
+                auto column_name = [&] {
+                    try {
+                        return column_name_type->from_string(name);
+                    } catch (marshal_exception&) {
+                        // #2597: Scylla < 2.0 writes names in serialized form, try to recover
+                        column_name_type->validate(to_bytes_view(name));
+                        return to_bytes(name);
+                    }
+                }();
+                builder.with_column_ordered(column_definition(std::move(column_name), std::move(validator), kind, component_index));
+            }
+
+            if (is_static_compact) {
+                builder.set_regular_column_name_type(db::schema_tables::parse_type(comparator));
+            }
+
+            if (td.has("gc_grace_seconds")) {
+                builder.set_gc_grace_seconds(td.get_as<int32_t>("gc_grace_seconds"));
+            }
+            if (td.has("min_compaction_threshold")) {
+                builder.set_min_compaction_threshold(td.get_as<int32_t>("min_compaction_threshold"));
+            }
+            if (td.has("max_compaction_threshold")) {
+                builder.set_max_compaction_threshold(td.get_as<int32_t>("max_compaction_threshold"));
+            }
+            if (td.has("comment")) {
+                builder.set_comment(td.get_as<sstring>("comment"));
+            }
+            if (td.has("memtable_flush_period_in_ms")) {
+                builder.set_memtable_flush_period(td.get_as<int32_t>("memtable_flush_period_in_ms"));
+            }
+            if (td.has("caching")) {
+                builder.set_caching_options(caching_options::from_sstring(td.get_as<sstring>("caching")));
+            }
+            if (td.has("default_time_to_live")) {
+                builder.set_default_time_to_live(gc_clock::duration(td.get_as<int32_t>("default_time_to_live")));
+            }
+            if (td.has("speculative_retry")) {
+                builder.set_speculative_retry(td.get_as<sstring>("speculative_retry"));
+            }
+            if (td.has("compaction_strategy_class")) {
+                auto strategy = td.get_as<sstring>("compaction_strategy_class");
+                try {
+                    builder.set_compaction_strategy(compaction::compaction_strategy::type(strategy));
+                } catch (const exceptions::configuration_exception& e) {
+                    // If compaction strategy class isn't supported, fallback to incremental.
+                    mlogger.warn("Falling back to incremental compaction strategy after the problem: {}", e.what());
+                    builder.set_compaction_strategy(compaction::compaction_strategy_type::incremental);
+                }
+            }
+            if (td.has("compaction_strategy_options")) {
+                sstring strategy_options_str = td.get_as<sstring>("compaction_strategy_options");
+                builder.set_compaction_strategy_options(rjson::parse_to_map<std::map<sstring, sstring>>(std::string_view(strategy_options_str)));
+            }
+            auto comp_param = td.get_as<sstring>("compression_parameters");
+            compression_parameters cp(rjson::parse_to_map<std::map<sstring, sstring>>(std::string_view(comp_param)));
+            builder.set_compressor_params(cp);
+
+            if (td.has("min_index_interval")) {
+                builder.set_min_index_interval(td.get_as<int32_t>("min_index_interval"));
+            } else if (td.has("index_interval")) { // compatibility
+                builder.set_min_index_interval(td.get_as<int32_t>("index_interval"));
+            }
+            if (td.has("max_index_interval")) {
+                builder.set_max_index_interval(td.get_as<int32_t>("max_index_interval"));
+            }
+            if (td.has("bloom_filter_fp_chance")) {
+                builder.set_bloom_filter_fp_chance(td.get_as<double>("bloom_filter_fp_chance"));
+            } else {
+                builder.set_bloom_filter_fp_chance(builder.get_bloom_filter_fp_chance());
+            }
+            if (td.has("dropped_columns")) {
+                auto map = td.get_map<sstring, int64_t>("dropped_columns");
+                for (auto&& e : map) {
+                    builder.without_column(e.first, api::timestamp_type(e.second));
+                };
+            }
+
+            // ignore version. we're transient
+            if (!triggers->empty()) {
+                throw unsupported_feature("triggers");
+            }
+
+            dst.tables.emplace_back(table{timestamp, builder.build() });
+        });
+    }
+
+    future<> read_tables(keyspace& dst) {
+        auto query = fmt_query("SELECT columnfamily_name, writeTime(type) AS timestamp FROM {}.{} WHERE keyspace_name = ?",
+                        db::system_keyspace::legacy::COLUMNFAMILIES);
+        return _qp.execute_internal(query, {dst.name}, cql3::query_processor::cache_internal::yes).then([this, &dst](result_set_type result) {
+            return parallel_for_each(*result, [this, &dst](row_type& row) {
+                return read_table(dst, row.get_as<sstring>("columnfamily_name"), row.get_as<time_point>("timestamp"));
+            }).finally([result] {});
+        });
+    }
+
+    future<time_point> read_type_timestamp(keyspace& dst, sstring type_name) {
+        // TODO: Unfortunately there is not a single REGULAR column in system.schema_usertypes, so annoyingly we cannot
+        // use the writeTime() CQL function, and must resort to a lower level.
+        // Origin digs up the actual cells of target partition and gets timestamp from there.
+        // We should do the same, but g-dam that's messy. Lets give back dung value for now.
+        return make_ready_future<time_point>(dst.timestamp);
+    }
+
+    future<> read_types(keyspace& dst) {
+        auto query = fmt_query("SELECT * FROM {}.{} WHERE keyspace_name = ?", db::system_keyspace::legacy::USERTYPES);
+        return _qp.execute_internal(query, {dst.name}, cql3::query_processor::cache_internal::yes).then([this, &dst](result_set_type result) {
+            return parallel_for_each(*result, [this, &dst](row_type& row) {
+                auto name = row.get_blob_unfragmented("type_name");
+                auto columns = row.get_list<bytes>("field_names");
+                auto types = row.get_list<sstring>("field_types");
+                std::vector<data_type> field_types;
+                for (auto&& value : types) {
+                    field_types.emplace_back(db::schema_tables::parse_type(value));
+                }
+                auto ut = user_type_impl::get_instance(dst.name, name, columns, field_types, false);
+                return read_type_timestamp(dst, value_cast<sstring>(utf8_type->deserialize(name))).then([ut = std::move(ut), &dst](time_point timestamp) {
+                    dst.types.emplace_back(type{timestamp, ut});
+                });
+            }).finally([result] {});
+        });
+    }
+
+    future<> read_functions(keyspace& dst) {
+        auto query = fmt_query("SELECT * FROM {}.{} WHERE keyspace_name = ?", db::system_keyspace::legacy::FUNCTIONS);
+        return _qp.execute_internal(query, {dst.name}, cql3::query_processor::cache_internal::yes).then([](result_set_type result) {
+            if (!result->empty()) {
+                throw unsupported_feature("functions");
+            }
+        });
+    }
+
+    future<> read_aggregates(keyspace& dst) {
+        auto query = fmt_query("SELECT * FROM {}.{} WHERE keyspace_name = ?", db::system_keyspace::legacy::AGGREGATES);
+        return _qp.execute_internal(query, {dst.name}, cql3::query_processor::cache_internal::yes).then([](result_set_type result) {
+            if (!result->empty()) {
+                throw unsupported_feature("aggregates");
+            }
+        });
+    }
+
+    future<keyspace> read_keyspace(sstring ks_name, bool durable_writes, sstring strategy_class, sstring strategy_options, time_point timestamp) {
+        auto map = rjson::parse_to_map<std::map<sstring, sstring>>(std::string_view(strategy_options));
+        map.emplace("class", std::move(strategy_class));
+        auto ks = ::make_lw_shared<keyspace>(keyspace{timestamp, std::move(ks_name), durable_writes, std::move(map) });
+
+        return read_tables(*ks).then([this, ks] {
+            //Collection<Type> types = readTypes(keyspaceName);
+            return read_types(*ks);
+        }).then([this, ks] {
+            return read_functions(*ks);
+        }).then([this, ks] {
+            return read_aggregates(*ks);
+        }).then([ks] {
+            return make_ready_future<keyspace>(std::move(*ks));
+        });
+    }
+
+    future<> read_all_keyspaces() {
+        static auto ks_filter = [](row_type& row) {
+            auto ks_name = row.get_as<sstring>("keyspace_name");
+            return ks_name != db::system_keyspace::NAME && ks_name != db::schema_tables::v3::NAME;
+        };
+
+        auto query = fmt_query("SELECT keyspace_name, durable_writes, strategy_options, strategy_class, writeTime(durable_writes) AS timestamp FROM {}.{}",
+                        db::system_keyspace::legacy::KEYSPACES);
+
+        return _qp.execute_internal(query, cql3::query_processor::cache_internal::yes).then([this](result_set_type result) {
+            auto i = boost::make_filter_iterator(ks_filter, result->begin(), result->end());
+            auto e = boost::make_filter_iterator(ks_filter, result->end(), result->end());
+            return parallel_for_each(i, e, [this](row_type& row) {
+                return read_keyspace(row.get_as<sstring>("keyspace_name")
+                                , row.get_as<bool>("durable_writes")
+                                , row.get_as<sstring>("strategy_class")
+                                , row.get_as<sstring>("strategy_options")
+                                , row.get_as<db_clock::time_point>("timestamp")
+                                ).then([this](keyspace ks) {
+                    _keyspaces.emplace_back(std::move(ks));
+                   });
+            }).finally([result] {});
+        });
+    }
+
+    future<> drop_legacy_tables() {
+        mlogger.info("Dropping legacy schema tables");
+        auto with_snapshot = !_keyspaces.empty();
+        for (const sstring& cfname : legacy_schema_tables) {
+            co_await replica::database::legacy_drop_table_on_all_shards(_db, _sys_ks, db::system_keyspace::NAME, cfname, with_snapshot);
+        }
+    }
+
+    future<> store_keyspaces_in_new_schema_tables() {
+        mlogger.info("Moving {} keyspaces from legacy schema tables to the new schema keyspace ({})",
+                        _keyspaces.size(), db::schema_tables::v3::NAME);
+
+        utils::chunked_vector<mutation> mutations;
+
+        for (auto& ks : _keyspaces) {
+            auto ksm = ::make_lw_shared<keyspace_metadata>(ks.name
+                            , ks.replication_params["class"] // TODO, make ksm like c3?
+                            , cql3::statements::property_definitions::to_extended_map(ks.replication_params)
+                            , std::nullopt
+                            , std::nullopt
+                            , ks.durable_writes);
+
+            // we want separate time stamps for tables/types, so cannot bulk them into the ksm.
+            for (auto&& m : db::schema_tables::make_create_keyspace_mutations(schema_features::full(), ksm, ks.timestamp.time_since_epoch().count(), false)) {
+                mutations.emplace_back(std::move(m));
+            }
+            for (auto& t : ks.tables) {
+                db::schema_tables::add_table_or_view_to_schema_mutation(t.metadata, t.timestamp.time_since_epoch().count(), true, mutations);
+            }
+            for (auto& t : ks.types) {
+                db::schema_tables::add_type_to_schema_mutation(t.metadata, t.timestamp.time_since_epoch().count(), mutations);
+            }
+        }
+        return _qp.proxy().mutate_locally(std::move(mutations), tracing::trace_state_ptr());
+    }
+
+    future<> flush_schemas() {
+        auto& db = _qp.db().real_database().container();
+        return replica::database::flush_tables_on_all_shards(db, db::schema_tables::all_table_infos(schema_features::full()));
+    }
+
+    future<> migrate() {
+        return read_all_keyspaces().then([this]() {
+            // write metadata to the new schema tables
+            return store_keyspaces_in_new_schema_tables()
+                                                .then(std::bind(&migrator::flush_schemas, this))
+                                                .then(std::bind(&migrator::drop_legacy_tables, this))
+                                                .then([] { mlogger.info("Completed migration of legacy schema tables"); });
+        });
+    }
+
+    sharded<service::storage_proxy>& _sp;
+    sharded<replica::database>& _db;
+    sharded<db::system_keyspace>& _sys_ks;
+    cql3::query_processor& _qp;
+    std::vector<keyspace> _keyspaces;
+};
+
+const std::unordered_set<sstring> migrator::legacy_schema_tables = {
+                db::system_keyspace::legacy::KEYSPACES,
+                db::system_keyspace::legacy::COLUMNFAMILIES,
+                db::system_keyspace::legacy::COLUMNS,
+                db::system_keyspace::legacy::TRIGGERS,
+                db::system_keyspace::legacy::USERTYPES,
+                db::system_keyspace::legacy::FUNCTIONS,
+                db::system_keyspace::legacy::AGGREGATES,
+};
+
+}
+}
+
+future<>
+db::legacy_schema_migrator::migrate(sharded<service::storage_proxy>& sp, sharded<replica::database>& db, sharded<db::system_keyspace>& sys_ks, cql3::query_processor& qp) {
+    return do_with(migrator(sp, db, sys_ks, qp), std::bind(&migrator::migrate, std::placeholders::_1));
+}
+

db/legacy_schema_migrator.hh

@@ -0,0 +1,37 @@
+/*
+ * Modified by ScyllaDB
+ * Copyright (C) 2017-present ScyllaDB
+ */
+
+/*
+ * SPDX-License-Identifier: (LicenseRef-ScyllaDB-Source-Available-1.0 and Apache-2.0)
+ */
+
+#pragma once
+
+#include <seastar/core/future.hh>
+#include <seastar/core/sharded.hh>
+
+#include "seastarx.hh"
+
+namespace replica {
+class database;
+}
+
+namespace cql3 {
+class query_processor;
+}
+
+namespace service {
+class storage_proxy;
+}
+
+namespace db {
+class system_keyspace;
+
+namespace legacy_schema_migrator {
+
+future<> migrate(sharded<service::storage_proxy>&, sharded<replica::database>& db, sharded<db::system_keyspace>& sys_ks, cql3::query_processor&);
+
+}
+}

db/partition_snapshot_row_cursor.hh

@@ -542,7 +542,6 @@ public:
     // Returns the range tombstone for the key range adjacent to the cursor's position from the side of smaller keys.
     // Excludes the range for the row itself. That information is returned by range_tombstone_for_row().
     // It's possible that range_tombstone() is empty and range_tombstone_for_row() is not empty.
-    // Note that this is different from the meaning of rows_entry::range_tombstone(), which includes the row itself.
     tombstone range_tombstone() const { return _range_tombstone; }
 
     // Can be called when cursor is pointing at a row.

db/row_cache.cc

@@ -1287,15 +1287,6 @@ row_cache::row_cache(schema_ptr s, snapshot_source src, cache_tracker& tracker,
     , _partitions(dht::raw_token_less_comparator{})
     , _underlying(src())
     , _snapshot_source(std::move(src))
-    , _update_section(abstract_formatter([this] (fmt::context& ctx) {
-        fmt::format_to(ctx.out(), "cache.update {}.{}", _schema->ks_name(), _schema->cf_name());
-    }))
-    , _populate_section(abstract_formatter([this] (fmt::context& ctx) {
-        fmt::format_to(ctx.out(), "cache.populate {}.{}", _schema->ks_name(), _schema->cf_name());
-    }))
-    , _read_section(abstract_formatter([this] (fmt::context& ctx) {
-        fmt::format_to(ctx.out(), "cache.read {}.{}", _schema->ks_name(), _schema->cf_name());
-    }))
 {
   try {
     with_allocator(_tracker.allocator(), [this, cont] {

db/schema_applier.cc

@@ -1121,7 +1121,7 @@ future<> schema_applier::commit() {
     // Run func first on shard 0
     // to allow "seeding" of the effective_replication_map
     // with a new e_r_m instance.
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     commit_on_shard(sharded_db.local());
     co_await sharded_db.invoke_on_others([this] (replica::database& db) {
         commit_on_shard(db);

db/schema_tables.cc

@@ -404,7 +404,10 @@ const std::unordered_set<table_id>& schema_tables_holding_schema_mutations() {
                 computed_columns(),
                 dropped_columns(),
                 indexes(),
-                scylla_tables()}) {
+                scylla_tables(),
+                db::system_keyspace::legacy::column_families(),
+                db::system_keyspace::legacy::columns(),
+                db::system_keyspace::legacy::triggers()}) {
             SCYLLA_ASSERT(s->clustering_key_size() > 0);
             auto&& first_column_name = s->clustering_column_at(0).name_as_text();
             SCYLLA_ASSERT(first_column_name == "table_name"
@@ -2837,6 +2840,26 @@ void check_no_legacy_secondary_index_mv_schema(replica::database& db, const view
 }
 
 
+namespace legacy {
+
+table_schema_version schema_mutations::digest() const {
+    md5_hasher h;
+    const db::schema_features no_features;
+    db::schema_tables::feed_hash_for_schema_digest(h, _columnfamilies, no_features);
+    db::schema_tables::feed_hash_for_schema_digest(h, _columns, no_features);
+    return table_schema_version(utils::UUID_gen::get_name_UUID(h.finalize()));
+}
+
+future<schema_mutations> read_table_mutations(sharded<service::storage_proxy>& proxy,
+    sstring keyspace_name, sstring table_name, schema_ptr s)
+{
+    mutation cf_m = co_await read_schema_partition_for_table(proxy, s, keyspace_name, table_name);
+    mutation col_m = co_await read_schema_partition_for_table(proxy, db::system_keyspace::legacy::columns(), keyspace_name, table_name);
+    co_return schema_mutations{std::move(cf_m), std::move(col_m)};
+}
+
+} // namespace legacy
+
 static auto GET_COLUMN_MAPPING_QUERY = format("SELECT column_name, clustering_order, column_name_bytes, kind, position, type FROM system.{} WHERE cf_id = ? AND schema_version = ?",
     db::schema_tables::SCYLLA_TABLE_SCHEMA_HISTORY);
 

db/schema_tables.hh

@@ -155,6 +155,24 @@ schema_ptr scylla_table_schema_history();
 const std::unordered_set<table_id>& schema_tables_holding_schema_mutations();
 }
 
+namespace legacy {
+
+class schema_mutations {
+    mutation _columnfamilies;
+    mutation _columns;
+public:
+    schema_mutations(mutation columnfamilies, mutation columns)
+        : _columnfamilies(std::move(columnfamilies))
+        , _columns(std::move(columns))
+    { }
+    table_schema_version digest() const;
+};
+
+future<schema_mutations> read_table_mutations(sharded<service::storage_proxy>& proxy,
+    sstring keyspace_name, sstring table_name, schema_ptr s);
+
+}
+
 struct qualified_name {
     sstring keyspace_name;
     sstring table_name;

db/size_estimates_virtual_reader.cc

@@ -187,7 +187,7 @@ static future<std::vector<token_range>> get_local_ranges(replica::database& db,
         auto ranges = db.get_token_metadata().get_primary_ranges_for(std::move(tokens));
         std::vector<token_range> local_ranges;
         auto to_bytes = [](const std::optional<dht::token_range::bound>& b) {
-            SCYLLA_ASSERT(b);
+            scylla_assert(b);
             return utf8_type->decompose(b->value().to_sstring());
         };
         // We merge the ranges to be compatible with how Cassandra shows it's size estimates table.

db/system_distributed_keyspace.cc

@@ -231,7 +231,7 @@ static schema_ptr get_current_service_levels(data_dictionary::database db) {
 }
 
 static schema_ptr get_updated_service_levels(data_dictionary::database db, bool workload_prioritization_enabled) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     auto schema = get_current_service_levels(db);
     schema_builder b(schema);
     for (const auto& col : new_service_levels_columns(workload_prioritization_enabled)) {

db/system_keyspace.cc

@@ -847,6 +847,8 @@ schema_ptr system_keyspace::corrupt_data() {
     return corrupt_data;
 }
 
+static constexpr auto schema_gc_grace = std::chrono::duration_cast<std::chrono::seconds>(days(7)).count();
+
 /*static*/ schema_ptr system_keyspace::scylla_local() {
     static thread_local auto scylla_local = [] {
         schema_builder builder(generate_legacy_id(NAME, SCYLLA_LOCAL), NAME, SCYLLA_LOCAL,
@@ -1358,6 +1360,289 @@ schema_ptr system_keyspace::role_permissions() {
     return schema;
 }
 
+schema_ptr system_keyspace::legacy::hints() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, HINTS), NAME, HINTS,
+        // partition key
+        {{"target_id", uuid_type}},
+        // clustering key
+        {{"hint_id", timeuuid_type}, {"message_version", int32_type}},
+        // regular columns
+        {{"mutation", bytes_type}},
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* hints awaiting delivery"
+       );
+       builder.set_gc_grace_seconds(0);
+       builder.set_compaction_strategy(compaction::compaction_strategy_type::incremental);
+       builder.set_compaction_strategy_options({{"enabled", "false"}});
+       builder.with(schema_builder::compact_storage::yes);
+       builder.with_hash_version();
+       return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::batchlog() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, BATCHLOG), NAME, BATCHLOG,
+        // partition key
+        {{"id", uuid_type}},
+        // clustering key
+        {},
+        // regular columns
+        {{"data", bytes_type}, {"version", int32_type}, {"written_at", timestamp_type}},
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* batchlog entries"
+       );
+       builder.set_gc_grace_seconds(0);
+       builder.set_compaction_strategy(compaction::compaction_strategy_type::incremental);
+       builder.set_compaction_strategy_options({{"min_threshold", "2"}});
+       builder.with(schema_builder::compact_storage::no);
+       builder.with_hash_version();
+       return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::keyspaces() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, KEYSPACES), NAME, KEYSPACES,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {},
+        // regular columns
+        {
+         {"durable_writes", boolean_type},
+         {"strategy_class", utf8_type},
+         {"strategy_options", utf8_type}
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* keyspace definitions"
+       );
+       builder.set_gc_grace_seconds(schema_gc_grace);
+       builder.with(schema_builder::compact_storage::yes);
+       builder.with_hash_version();
+       return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::column_families() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, COLUMNFAMILIES), NAME, COLUMNFAMILIES,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"columnfamily_name", utf8_type}},
+        // regular columns
+        {
+         {"bloom_filter_fp_chance", double_type},
+         {"caching", utf8_type},
+         {"cf_id", uuid_type},
+         {"comment", utf8_type},
+         {"compaction_strategy_class", utf8_type},
+         {"compaction_strategy_options", utf8_type},
+         {"comparator", utf8_type},
+         {"compression_parameters", utf8_type},
+         {"default_time_to_live", int32_type},
+         {"default_validator", utf8_type},
+         {"dropped_columns",  map_type_impl::get_instance(utf8_type, long_type, true)},
+         {"gc_grace_seconds", int32_type},
+         {"is_dense", boolean_type},
+         {"key_validator", utf8_type},
+         {"max_compaction_threshold", int32_type},
+         {"max_index_interval", int32_type},
+         {"memtable_flush_period_in_ms", int32_type},
+         {"min_compaction_threshold", int32_type},
+         {"min_index_interval", int32_type},
+         {"speculative_retry", utf8_type},
+         {"subcomparator", utf8_type},
+         {"type", utf8_type},
+         // The following 4 columns are only present up until 2.1.8 tables
+         {"key_aliases", utf8_type},
+         {"value_alias", utf8_type},
+         {"column_aliases", utf8_type},
+         {"index_interval", int32_type},},
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* table definitions"
+       );
+       builder.set_gc_grace_seconds(schema_gc_grace);
+       builder.with(schema_builder::compact_storage::no);
+       builder.with_hash_version();
+       return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::columns() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, COLUMNS), NAME, COLUMNS,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"columnfamily_name", utf8_type}, {"column_name", utf8_type}},
+        // regular columns
+        {
+            {"component_index", int32_type},
+            {"index_name", utf8_type},
+            {"index_options", utf8_type},
+            {"index_type", utf8_type},
+            {"type", utf8_type},
+            {"validator", utf8_type},
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "column definitions"
+        );
+        builder.set_gc_grace_seconds(schema_gc_grace);
+        builder.with(schema_builder::compact_storage::no);
+        builder.with_hash_version();
+        return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::triggers() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, TRIGGERS), NAME, TRIGGERS,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"columnfamily_name", utf8_type}, {"trigger_name", utf8_type}},
+        // regular columns
+        {
+            {"trigger_options",  map_type_impl::get_instance(utf8_type, utf8_type, true)},
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "trigger definitions"
+        );
+        builder.set_gc_grace_seconds(schema_gc_grace);
+        builder.with(schema_builder::compact_storage::no);
+        builder.with_hash_version();
+        return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::usertypes() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, USERTYPES), NAME, USERTYPES,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"type_name", utf8_type}},
+        // regular columns
+        {
+            {"field_names", list_type_impl::get_instance(utf8_type, true)},
+            {"field_types", list_type_impl::get_instance(utf8_type, true)},
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "user defined type definitions"
+        );
+        builder.set_gc_grace_seconds(schema_gc_grace);
+        builder.with(schema_builder::compact_storage::no);
+        builder.with_hash_version();
+        return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::functions() {
+    /**
+     * Note: we have our own "legacy" version of this table (in schema_tables),
+     * but it is (afaik) not used, and differs slightly from the origin one.
+     * This is based on the origin schema, since we're more likely to encounter
+     * installations of that to migrate, rather than our own (if we dont use the table).
+     */
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, FUNCTIONS), NAME, FUNCTIONS,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"function_name", utf8_type},{"signature", list_type_impl::get_instance(utf8_type, false)}},
+        // regular columns
+        {
+            {"argument_names", list_type_impl::get_instance(utf8_type, true)},
+            {"argument_types", list_type_impl::get_instance(utf8_type, true)},
+            {"body", utf8_type},
+            {"language", utf8_type},
+            {"return_type", utf8_type},
+            {"called_on_null_input", boolean_type},
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* user defined type definitions"
+        );
+        builder.set_gc_grace_seconds(schema_gc_grace);
+        builder.with(schema_builder::compact_storage::no);
+        builder.with_hash_version();
+        return builder.build();
+    }();
+    return schema;
+}
+
+schema_ptr system_keyspace::legacy::aggregates() {
+    static thread_local auto schema = [] {
+        schema_builder builder(generate_legacy_id(NAME, AGGREGATES), NAME, AGGREGATES,
+        // partition key
+        {{"keyspace_name", utf8_type}},
+        // clustering key
+        {{"aggregate_name", utf8_type},{"signature", list_type_impl::get_instance(utf8_type, false)}},
+        // regular columns
+        {
+            {"argument_types", list_type_impl::get_instance(utf8_type, true)},
+            {"final_func", utf8_type},
+            {"initcond", bytes_type},
+            {"return_type", utf8_type},
+            {"state_func", utf8_type},
+            {"state_type", utf8_type},
+        },
+        // static columns
+        {},
+        // regular column name type
+        utf8_type,
+        // comment
+        "*DEPRECATED* user defined aggregate definition"
+        );
+        builder.set_gc_grace_seconds(schema_gc_grace);
+        builder.with(schema_builder::compact_storage::no);
+        builder.with_hash_version();
+        return builder.build();
+    }();
+    return schema;
+}
+
 schema_ptr system_keyspace::dicts() {
     static thread_local auto schema = [] {
         auto id = generate_legacy_id(NAME, DICTS);
@@ -2330,6 +2615,13 @@ std::vector<schema_ptr> system_keyspace::all_tables(const db::config& cfg) {
     if (cfg.check_experimental(db::experimental_features_t::feature::KEYSPACE_STORAGE_OPTIONS)) {
         r.insert(r.end(), {sstables_registry()});
     }
+    // legacy schema
+    r.insert(r.end(), {
+                    // TODO: once we migrate hints/batchlog and add converter
+                    // legacy::hints(), legacy::batchlog(),
+                    legacy::keyspaces(), legacy::column_families(),
+                    legacy::columns(), legacy::triggers(), legacy::usertypes(),
+                    legacy::functions(), legacy::aggregates(), });
 
     return r;
 }

db/system_keyspace.hh

@@ -241,6 +241,28 @@ public:
         static schema_ptr cdc_local();
     };
 
+    struct legacy {
+        static constexpr auto HINTS = "hints";
+        static constexpr auto BATCHLOG = "batchlog";
+        static constexpr auto KEYSPACES = "schema_keyspaces";
+        static constexpr auto COLUMNFAMILIES = "schema_columnfamilies";
+        static constexpr auto COLUMNS = "schema_columns";
+        static constexpr auto TRIGGERS = "schema_triggers";
+        static constexpr auto USERTYPES = "schema_usertypes";
+        static constexpr auto FUNCTIONS = "schema_functions";
+        static constexpr auto AGGREGATES = "schema_aggregates";
+
+        static schema_ptr keyspaces();
+        static schema_ptr column_families();
+        static schema_ptr columns();
+        static schema_ptr triggers();
+        static schema_ptr usertypes();
+        static schema_ptr functions();
+        static schema_ptr aggregates();
+        static schema_ptr hints();
+        static schema_ptr batchlog();
+    };
+
     // Partition estimates for a given range of tokens.
     struct range_estimates {
         schema_ptr schema;

db/view/row_locking.cc

@@ -153,14 +153,14 @@ row_locker::unlock(const dht::decorated_key* pk, bool partition_exclusive,
             mylog.error("column_family::local_base_lock_holder::~local_base_lock_holder() can't find lock for partition", *pk);
             return;
         }
-        SCYLLA_ASSERT(&pli->first == pk);
+        scylla_assert(&pli->first == pk);
         if (cpk) {
             auto rli = pli->second._row_locks.find(*cpk);
             if (rli == pli->second._row_locks.end()) {
                 mylog.error("column_family::local_base_lock_holder::~local_base_lock_holder() can't find lock for row", *cpk);
                 return;
             }
-            SCYLLA_ASSERT(&rli->first == cpk);
+            scylla_assert(&rli->first == cpk);
             mylog.debug("releasing {} lock for row {} in partition {}", (row_exclusive ? "exclusive" : "shared"), *cpk, *pk);
             auto& lock = rli->second;
             if (row_exclusive) {

db/view/view.cc

@@ -1744,115 +1744,6 @@ bool should_generate_view_updates_on_this_shard(const schema_ptr& base, const lo
         && std::ranges::contains(shards, this_shard_id());
 }
 
-static endpoints_to_update get_view_natural_endpoint_vnodes(
-        locator::host_id me,
-        std::vector<std::reference_wrapper<const locator::node>> base_nodes,
-        std::vector<std::reference_wrapper<const locator::node>> view_nodes,
-        locator::endpoint_dc_rack my_location,
-        const locator::network_topology_strategy* network_topology,
-        replica::cf_stats& cf_stats) {
-    using node_vector = std::vector<std::reference_wrapper<const locator::node>>;
-    node_vector base_endpoints, view_endpoints;
-    auto& my_datacenter = my_location.dc;
-
-    auto process_candidate = [&] (node_vector& nodes, std::reference_wrapper<const locator::node> node) {
-        if (!network_topology || node.get().dc() == my_datacenter) {
-            nodes.emplace_back(node);
-        }
-    };
-
-    for (auto&& base_node : base_nodes) {
-        process_candidate(base_endpoints, base_node);
-    }
-
-    for (auto&& view_node : view_nodes) {
-        auto it = std::ranges::find(base_endpoints, view_node.get().host_id(), std::mem_fn(&locator::node::host_id));
-        // If this base replica is also one of the view replicas, we use
-        // ourselves as the view replica.
-        // We don't return an extra endpoint, as it's only needed when
-        // using tablets (so !use_legacy_self_pairing)
-        if (view_node.get().host_id() == me && it != base_endpoints.end()) {
-            return {.natural_endpoint = me};
-        }
-
-        // We have to remove any endpoint which is shared between the base
-        // and the view, as it will select itself and throw off the counts
-        // otherwise.
-        if (it != base_endpoints.end()) {
-            base_endpoints.erase(it);
-        } else if (!network_topology || view_node.get().dc() == my_datacenter) {
-            view_endpoints.push_back(view_node);
-        }
-    }
-
-    auto base_it = std::ranges::find(base_endpoints, me, std::mem_fn(&locator::node::host_id));
-    if (base_it == base_endpoints.end()) {
-        // This node is not a base replica of this key, so we return empty
-        // FIXME: This case shouldn't happen, and if it happens, a view update
-        // would be lost.
-        ++cf_stats.total_view_updates_on_wrong_node;
-        vlogger.warn("Could not find {} in base_endpoints={}", me,
-                base_endpoints | std::views::transform(std::mem_fn(&locator::node::host_id)));
-        return {};
-    }
-    size_t idx = base_it - base_endpoints.begin();
-    return {.natural_endpoint = view_endpoints[idx].get().host_id()};
-}
-
-static std::optional<locator::host_id> get_unpaired_view_endpoint(
-        std::vector<std::reference_wrapper<const locator::node>> base_nodes,
-        std::vector<std::reference_wrapper<const locator::node>> view_nodes,
-        replica::cf_stats& cf_stats) {
-    std::unordered_set<locator::endpoint_dc_rack> base_dc_racks;
-    for (auto&& base_node : base_nodes) {
-        if (base_dc_racks.contains(base_node.get().dc_rack())) {
-            // We can't do rack-aware pairing if there are multiple replicas in the same rack.
-            ++cf_stats.total_view_updates_failed_pairing;
-            vlogger.warn("Can't perform base-view pairing in this topology. There are multiple base table replicas in the same dc/rack({}/{}):",
-                    base_node.get().dc(), base_node.get().rack());
-            return std::nullopt;
-        }
-        base_dc_racks.insert(base_node.get().dc_rack());
-    }
-
-    std::unordered_set<locator::endpoint_dc_rack> paired_view_dc_racks;
-    std::unordered_map<locator::endpoint_dc_rack, locator::host_id> unpaired_view_dc_rack_replicas;
-    for (auto&& view_node : view_nodes) {
-        if (paired_view_dc_racks.contains(view_node.get().dc_rack()) || unpaired_view_dc_rack_replicas.contains(view_node.get().dc_rack())) {
-            // We can't do rack-aware pairing if there are multiple replicas in the same rack.
-            ++cf_stats.total_view_updates_failed_pairing;
-            vlogger.warn("Can't perform base-view pairing in this topology. There are multiple view table replicas in the same dc/rack({}/{}):",
-                    view_node.get().dc(), view_node.get().rack());
-            return std::nullopt;
-        }
-        // Track unpaired replicas in both sets
-        if (base_dc_racks.contains(view_node.get().dc_rack())) {
-            paired_view_dc_racks.insert(view_node.get().dc_rack());
-        } else {
-            unpaired_view_dc_rack_replicas.insert({view_node.get().dc_rack(), view_node.get().host_id()});
-        }
-    }
-
-    if (unpaired_view_dc_rack_replicas.size() > 0) {
-        // There are view replicas that can't be paired with any base replica
-        // This can happen as a result of an RF change when the view replica finishes streaming
-        // before the base replica.
-        // Because of this, a view replica might not get paired with any base replica, so we need
-        // to send an additional update to it.
-        ++cf_stats.total_view_updates_due_to_replica_count_mismatch;
-        auto extra_replica = unpaired_view_dc_rack_replicas.begin()->second;
-        unpaired_view_dc_rack_replicas.erase(unpaired_view_dc_rack_replicas.begin());
-        if (unpaired_view_dc_rack_replicas.size() > 0) {
-            // We only expect one extra replica to appear due to an RF change. If there's more, that's an error,
-            // but we'll still perform updates to the paired and last replicas to minimize degradation.
-            vlogger.warn("There are too many view endpoints for base-view pairing. View updates may get lost on view_endpoints={}",
-                    unpaired_view_dc_rack_replicas | std::views::values);
-        }
-        return extra_replica;
-    }
-    return std::nullopt;
-}
-
 // Calculate the node ("natural endpoint") to which this node should send
 // a view update.
 //
@@ -1865,19 +1756,29 @@ static std::optional<locator::host_id> get_unpaired_view_endpoint(
 // of this function is to find, assuming that this node is one of the base
 // replicas for a given partition, the paired view replica.
 //
-// When using vnodes, we have an optimization called "self-pairing" - if a single
-// node is both a base replica and a view replica for a write, the pairing is
-// modified so that this node sends the update to itself and this node is removed
-// from the lists of nodes paired by index. This self-pairing optimization can
-// cause the pairing to change after view ranges are moved between nodes.
+// In the past, we used an optimization called "self-pairing" that if a single
+// node was both a base replica and a view replica for a write, the pairing is
+// modified so that this node would send the update to itself. This self-
+// pairing optimization could cause the pairing to change after view ranges
+// are moved between nodes, so currently we only use it if
+// use_legacy_self_pairing is set to true. When using tablets - where range
+// movements are common - it is strongly recommended to set it to false.
 //
 // If the keyspace's replication strategy is a NetworkTopologyStrategy,
 // we pair only nodes in the same datacenter.
 //
-// If the table uses tablets, then pairing is rack-aware. In this case, in each
-// rack where we have a base replica there is also one replica of each view tablet.
-// Therefore, the base replicas are naturally paired with the view replicas that
-// are in the same rack.
+// When use_legacy_self_pairing is enabled, if one of the base replicas
+// also happens to be a view replica, it is paired with itself
+// (with the other nodes paired by order in the list
+// after taking this node out).
+//
+// If the table uses tablets and the replication strategy is NetworkTopologyStrategy
+// and the replication factor in the node's datacenter is a multiple of the number
+// of racks in the datacenter, then pairing is rack-aware.  In this case,
+// all racks have the same number of replicas, and those are never migrated
+// outside their racks. Therefore, the base replicas are naturally paired with the
+// view replicas that are in the same rack, based on the ordinal position.
+// Note that typically, there is a single replica per rack and pairing is trivial.
 //
 // If the assumption that the given base token belongs to this replica
 // does not hold, we return an empty optional.
@@ -1905,12 +1806,19 @@ endpoints_to_update get_view_natural_endpoint(
         const locator::abstract_replication_strategy& replication_strategy,
         const dht::token& base_token,
         const dht::token& view_token,
-        bool use_tablets,
+        bool use_legacy_self_pairing,
+        bool use_tablets_rack_aware_view_pairing,
         replica::cf_stats& cf_stats) {
     auto& topology = base_erm->get_token_metadata_ptr()->get_topology();
     auto& view_topology = view_erm->get_token_metadata_ptr()->get_topology();
     auto& my_location = topology.get_location(me);
+    auto& my_datacenter = my_location.dc;
     auto* network_topology = dynamic_cast<const locator::network_topology_strategy*>(&replication_strategy);
+    auto rack_aware_pairing = use_tablets_rack_aware_view_pairing && network_topology;
+    bool simple_rack_aware_pairing = false;
+    using node_vector = std::vector<std::reference_wrapper<const locator::node>>;
+    node_vector orig_base_endpoints, orig_view_endpoints;
+    node_vector base_endpoints, view_endpoints;
 
     auto resolve = [&] (const locator::topology& topology, const locator::host_id& ep, bool is_view) -> const locator::node& {
         if (auto* np = topology.find_node(ep)) {
@@ -1921,7 +1829,6 @@ endpoints_to_update get_view_natural_endpoint(
 
     // We need to use get_replicas() for pairing to be stable in case base or view tablet
     // is rebuilding a replica which has left the ring. get_natural_endpoints() filters such replicas.
-    using node_vector = std::vector<std::reference_wrapper<const locator::node>>;
     auto base_nodes = base_erm->get_replicas(base_token) | std::views::transform([&] (const locator::host_id& ep) -> const locator::node& {
         return resolve(topology, ep, false);
     }) | std::ranges::to<node_vector>();
@@ -1945,43 +1852,231 @@ endpoints_to_update get_view_natural_endpoint(
                 // note that the recursive call will not recurse again because leaving_base is in base_nodes.
                 auto leaving_base = it->get().host_id();
                 return get_view_natural_endpoint(leaving_base, base_erm, view_erm, replication_strategy, base_token,
-                        view_token, use_tablets, cf_stats);
+                        view_token, use_legacy_self_pairing, use_tablets_rack_aware_view_pairing, cf_stats);
             }
         }
     }
 
-    if (!use_tablets) {
-        return get_view_natural_endpoint_vnodes(
-                me,
-                base_nodes,
-                view_nodes,
-                my_location,
-                network_topology,
-                cf_stats);
+    std::function<bool(const locator::node&)> is_candidate;
+    if (network_topology) {
+        is_candidate = [&] (const locator::node& node) { return node.dc() == my_datacenter; };
+    } else {
+        is_candidate = [&] (const locator::node&) { return true; };
+    }
+    auto process_candidate = [&] (node_vector& nodes, std::reference_wrapper<const locator::node> node) {
+        if (is_candidate(node)) {
+            nodes.emplace_back(node);
+        }
+    };
+
+    for (auto&& base_node : base_nodes) {
+        process_candidate(base_endpoints, base_node);
     }
 
-    std::optional<locator::host_id> paired_replica;
-    for (auto&& view_node : view_nodes) {
-        if (view_node.get().dc_rack() == my_location) {
-            paired_replica = view_node.get().host_id();
-            break;
+    if (use_legacy_self_pairing) {
+        for (auto&& view_node : view_nodes) {
+            auto it = std::ranges::find(base_endpoints, view_node.get().host_id(), std::mem_fn(&locator::node::host_id));
+            // If this base replica is also one of the view replicas, we use
+            // ourselves as the view replica.
+            // We don't return an extra endpoint, as it's only needed when
+            // using tablets (so !use_legacy_self_pairing)
+            if (view_node.get().host_id() == me && it != base_endpoints.end()) {
+                return {.natural_endpoint = me};
+            }
+
+            // We have to remove any endpoint which is shared between the base
+            // and the view, as it will select itself and throw off the counts
+            // otherwise.
+            if (it != base_endpoints.end()) {
+                base_endpoints.erase(it);
+            } else if (is_candidate(view_node)) {
+                view_endpoints.push_back(view_node);
+            }
+        }
+    } else {
+        for (auto&& view_node : view_nodes) {
+            process_candidate(view_endpoints, view_node);
         }
     }
-    if (paired_replica && base_nodes.size() == view_nodes.size()) {
-        // We don't need to find any extra replicas, so we can return early
-        return {.natural_endpoint = paired_replica};
+
+    // Try optimizing for simple rack-aware pairing
+    // If the numbers of base and view replica differ, that means an RF change is taking place
+    // and we can't use simple rack-aware pairing.
+    if (rack_aware_pairing && base_endpoints.size() == view_endpoints.size()) {
+        auto dc_rf = network_topology->get_replication_factor(my_datacenter);
+        const auto& racks = topology.get_datacenter_rack_nodes().at(my_datacenter);
+        // Simple rack-aware pairing is possible when the datacenter replication factor
+        // is a multiple of the number of racks in the datacenter.
+        if (dc_rf % racks.size() == 0) {
+            simple_rack_aware_pairing = true;
+            size_t rack_rf = dc_rf / racks.size();
+            // If any rack doesn't have enough nodes to satisfy the per-rack rf
+            // simple rack-aware pairing is disabled.
+            for (const auto& [rack, nodes] : racks) {
+                if (nodes.size() < rack_rf) {
+                    simple_rack_aware_pairing = false;
+                    break;
+                }
+            }
+        }
+        if (dc_rf != base_endpoints.size()) {
+            // If the datacenter replication factor is not equal to the number of base replicas,
+            // we're in progress of a RF change and we can't use simple rack-aware pairing.
+            simple_rack_aware_pairing = false;
+        }
+        if (simple_rack_aware_pairing) {
+            std::erase_if(base_endpoints, [&] (const locator::node& node) { return node.dc_rack() != my_location; });
+            std::erase_if(view_endpoints, [&] (const locator::node& node) { return node.dc_rack() != my_location; });
+        }
     }
-    if (!paired_replica) {
-        // We couldn't find any view replica in our rack
+
+    orig_base_endpoints = base_endpoints;
+    orig_view_endpoints = view_endpoints;
+
+    // For the complex rack_aware_pairing case, nodes are already filtered by datacenter
+    // Use best-match, for the minimum number of base and view replicas in each rack,
+    // and ordinal match for the rest.
+    std::optional<std::reference_wrapper<const locator::node>> paired_replica;
+    if (rack_aware_pairing && !simple_rack_aware_pairing) {
+        struct indexed_replica {
+            size_t idx;
+            std::reference_wrapper<const locator::node> node;
+        };
+        std::unordered_map<sstring, std::vector<indexed_replica>> base_racks, view_racks;
+
+        // First, index all replicas by rack
+        auto index_replica_set = [] (std::unordered_map<sstring, std::vector<indexed_replica>>& racks, const node_vector& replicas) {
+            size_t idx = 0;
+            for (const auto& r: replicas) {
+                racks[r.get().rack()].emplace_back(idx++, r);
+            }
+        };
+        index_replica_set(base_racks, base_endpoints);
+        index_replica_set(view_racks, view_endpoints);
+
+        // Try optimistically pairing `me` first
+        const auto& my_base_replicas = base_racks[my_location.rack];
+        auto base_it = std::ranges::find(my_base_replicas, me, [] (const indexed_replica& ir) { return ir.node.get().host_id(); });
+        if (base_it == my_base_replicas.end()) {
+            return {};
+        }
+        const auto& my_view_replicas = view_racks[my_location.rack];
+        size_t idx = base_it - my_base_replicas.begin();
+        if (idx < my_view_replicas.size()) {
+            if (orig_view_endpoints.size() <= orig_base_endpoints.size()) {
+                return {.natural_endpoint = my_view_replicas[idx].node.get().host_id()};
+            } else {
+                // If the number of view replicas is larger than the number of base replicas,
+                // we need to find the unpaired view replica, so we can't return yet.
+                paired_replica = my_view_replicas[idx].node;
+            }
+        }
+
+        // Collect all unpaired base and view replicas,
+        // where the number of replicas in the base rack is different than the respective view rack
+        std::vector<indexed_replica> unpaired_base_replicas, unpaired_view_replicas;
+        for (const auto& [rack, base_replicas] : base_racks) {
+            const auto& view_replicas = view_racks[rack];
+            for (auto i = view_replicas.size(); i < base_replicas.size(); ++i) {
+                unpaired_base_replicas.emplace_back(base_replicas[i]);
+            }
+        }
+        for (const auto& [rack, view_replicas] : view_racks) {
+            const auto& base_replicas = base_racks[rack];
+            for (auto i = base_replicas.size(); i < view_replicas.size(); ++i) {
+                unpaired_view_replicas.emplace_back(view_replicas[i]);
+            }
+        }
+
+        // Sort by the original ordinality, and copy the sorted results
+        // back into {base,view}_endpoints, for backward compatible processing below.
+        std::ranges::sort(unpaired_base_replicas, std::less(), std::mem_fn(&indexed_replica::idx));
+        base_endpoints.clear();
+        std::ranges::transform(unpaired_base_replicas, std::back_inserter(base_endpoints), std::mem_fn(&indexed_replica::node));
+
+        std::ranges::sort(unpaired_view_replicas, std::less(), std::mem_fn(&indexed_replica::idx));
+        view_endpoints.clear();
+        std::ranges::transform(unpaired_view_replicas, std::back_inserter(view_endpoints), std::mem_fn(&indexed_replica::node));
+    }
+
+    auto base_it = std::ranges::find(base_endpoints, me, std::mem_fn(&locator::node::host_id));
+    if (!paired_replica && base_it == base_endpoints.end()) {
+        // This node is not a base replica of this key, so we return empty
+        // FIXME: This case shouldn't happen, and if it happens, a view update
+        // would be lost.
+        ++cf_stats.total_view_updates_on_wrong_node;
+        vlogger.warn("Could not find {} in base_endpoints={}", me,
+                orig_base_endpoints | std::views::transform(std::mem_fn(&locator::node::host_id)));
+        return {};
+    }
+    size_t idx = base_it - base_endpoints.begin();
+    std::optional<std::reference_wrapper<const locator::node>> no_pairing_replica;
+    if (!paired_replica && idx >= view_endpoints.size()) {
+        // There are fewer view replicas than base replicas
+        // FIXME: This might still happen when reducing replication factor with tablets,
+        // see https://github.com/scylladb/scylladb/issues/21492
         ++cf_stats.total_view_updates_failed_pairing;
-        vlogger.warn("Could not find a view replica in the same rack as base replica {} for base_endpoints={} view_endpoints={}", 
-                me,
-                base_nodes | std::views::transform(std::mem_fn(&locator::node::host_id)),
-                view_nodes | std::views::transform(std::mem_fn(&locator::node::host_id)));
+        vlogger.warn("Could not pair {}: rack_aware={} base_endpoints={} view_endpoints={}", me,
+                rack_aware_pairing ? (simple_rack_aware_pairing ? "simple" : "complex") : "none",
+                orig_base_endpoints | std::views::transform(std::mem_fn(&locator::node::host_id)),
+                orig_view_endpoints | std::views::transform(std::mem_fn(&locator::node::host_id)));
+        return {};
+    } else if (base_endpoints.size() < view_endpoints.size()) {
+        // There are fewer base replicas than view replicas.
+        // This can happen as a result of an RF change when the view replica finishes streaming
+        // before the base replica.
+        // Because of this, a view replica might not get paired with any base replica, so we need
+        // to send an additional update to it.
+        ++cf_stats.total_view_updates_due_to_replica_count_mismatch;
+        no_pairing_replica = view_endpoints.back();
+        if (base_endpoints.size() < view_endpoints.size() - 1) {
+            // We only expect one extra replica to appear due to an RF change. If there's more, that's an error,
+            // but we'll still perform updates to the paired and last replicas to minimize degradation.
+            vlogger.warn("There are too many view endpoints for base-view pairing. View updates may get lost on view_endpoints={}",
+                    std::span(view_endpoints.begin() + base_endpoints.size(), view_endpoints.end() - 1) | std::views::transform(std::mem_fn(&locator::node::host_id)));
+        }
     }
-    std::optional<locator::host_id> no_pairing_replica = get_unpaired_view_endpoint(base_nodes, view_nodes, cf_stats);
-    return {.natural_endpoint = paired_replica,
-            .endpoint_with_no_pairing = no_pairing_replica};
+
+    if (!paired_replica) {
+        paired_replica = view_endpoints[idx];
+    }
+    if (!no_pairing_replica && base_nodes.size() < view_nodes.size()) {
+        // This can happen when the view replica with no pairing is in another DC.
+        // We need to send an update to it if there are no base replicas in that DC yet,
+        // as it won't receive updates otherwise.
+        std::unordered_set<sstring> dcs_with_base_replicas;
+        for (const auto& base_node : base_nodes) {
+            dcs_with_base_replicas.insert(base_node.get().dc());
+        }
+        for (const auto& view_node : view_nodes) {
+            if (!dcs_with_base_replicas.contains(view_node.get().dc())) {
+                ++cf_stats.total_view_updates_due_to_replica_count_mismatch;
+                no_pairing_replica = view_node;
+                break;
+            }
+        }
+    }
+    // https://github.com/scylladb/scylladb/issues/19439
+    // With tablets, a node being replaced might transition to "left" state
+    // but still be kept as a replica.
+    // As of writing this hints are not prepared to handle nodes that are left
+    // but are still replicas. Therefore, there is no other sensible option
+    // right now but to give up attempt to send the update or write a hint
+    // to the paired, permanently down replica.
+    // We use the same workaround for the extra replica.
+    auto return_host_id_if_not_left = [] (const auto& replica) -> std::optional<locator::host_id> {
+        if (!replica) {
+            return std::nullopt;
+        }
+        const auto& node = replica->get();
+        if (!node.left()) {
+            return node.host_id();
+        } else {
+            return std::nullopt;
+        }
+    };
+    return {.natural_endpoint = return_host_id_if_not_left(paired_replica),
+            .endpoint_with_no_pairing = return_host_id_if_not_left(no_pairing_replica)};
 }
 
 static future<> apply_to_remote_endpoints(service::storage_proxy& proxy, locator::effective_replication_map_ptr ermp,
@@ -2041,6 +2136,12 @@ future<> view_update_generator::mutate_MV(
 {
     auto& ks = _db.find_keyspace(base->ks_name());
     auto& replication = ks.get_replication_strategy();
+    // We set legacy self-pairing for old vnode-based tables (for backward
+    // compatibility), and unset it for tablets - where range movements
+    // are more frequent and backward compatibility is less important.
+    // TODO: Maybe allow users to set use_legacy_self_pairing explicitly
+    // on a view, like we have the synchronous_updates_flag.
+    bool use_legacy_self_pairing = !ks.uses_tablets();
     std::unordered_map<table_id, locator::effective_replication_map_ptr> erms;
     auto get_erm = [&] (table_id id) {
         auto it = erms.find(id);
@@ -2053,6 +2154,10 @@ future<> view_update_generator::mutate_MV(
     for (const auto& mut : view_updates) {
         (void)get_erm(mut.s->id());
     }
+    // Enable rack-aware view updates pairing for tablets
+    // when the cluster feature is enabled so that all replicas agree
+    // on the pairing algorithm.
+    bool use_tablets_rack_aware_view_pairing = _db.features().tablet_rack_aware_view_pairing && ks.uses_tablets();
     auto me = base_ermp->get_topology().my_host_id();
     static constexpr size_t max_concurrent_updates = 128;
     co_await utils::get_local_injector().inject("delay_before_get_view_natural_endpoint", 8000ms);
@@ -2060,7 +2165,7 @@ future<> view_update_generator::mutate_MV(
         auto view_token = dht::get_token(*mut.s, mut.fm.key());
         auto view_ermp = erms.at(mut.s->id());
         auto [target_endpoint, no_pairing_endpoint] = get_view_natural_endpoint(me, base_ermp, view_ermp, replication, base_token, view_token,
-                ks.uses_tablets(), cf_stats);
+                use_legacy_self_pairing, use_tablets_rack_aware_view_pairing, cf_stats);
         auto remote_endpoints = view_ermp->get_pending_replicas(view_token);
         auto memory_units = seastar::make_lw_shared<db::timeout_semaphore_units>(pending_view_update_memory_units.split(memory_usage_of(mut)));
         if (no_pairing_endpoint) {

db/view/view.hh

@@ -305,7 +305,8 @@ endpoints_to_update get_view_natural_endpoint(
     const locator::abstract_replication_strategy& replication_strategy,
     const dht::token& base_token,
     const dht::token& view_token,
-    bool use_tablets,
+    bool use_legacy_self_pairing,
+    bool use_tablets_basic_rack_aware_view_pairing,
     replica::cf_stats& cf_stats);
 
 /// Verify that the provided keyspace is eligible for storing materialized views.

dist/common/sysconfig/scylla-node-exporter

@@ -1 +1 @@
-SCYLLA_NODE_EXPORTER_ARGS="--collector.interrupts --collector.ethtool.metrics-include='(bw_in_allowance_exceeded|bw_out_allowance_exceeded|conntrack_allowance_exceeded|conntrack_allowance_available|linklocal_allowance_exceeded)' --collector.ethtool --no-collector.hwmon --no-collector.bcache --no-collector.btrfs --no-collector.fibrechannel --no-collector.infiniband --no-collector.ipvs --no-collector.nfs --no-collector.nfsd --no-collector.powersupplyclass --no-collector.rapl --no-collector.tapestats --no-collector.thermal_zone --no-collector.udp_queues --no-collector.zfs"
+SCYLLA_NODE_EXPORTER_ARGS="--collector.interrupts --no-collector.hwmon --no-collector.bcache --no-collector.btrfs --no-collector.fibrechannel --no-collector.infiniband --no-collector.ipvs --no-collector.nfs --no-collector.nfsd --no-collector.powersupplyclass --no-collector.rapl --no-collector.tapestats --no-collector.thermal_zone --no-collector.udp_queues --no-collector.zfs"

docs/_ext/scylladb_metrics.py

@@ -41,8 +41,6 @@ class MetricsProcessor:
                 # Get metrics from the file
                 try:
                     metrics_file = metrics.get_metrics_from_file(relative_path, "scylla_", metrics_info, strict=strict)
-                except SystemExit:
-                    pass
                 finally:
                     os.chdir(old_cwd)
                 if metrics_file:

docs/dev/scylla_assert_conversion.md

@@ -0,0 +1,198 @@
+# SCYLLA_ASSERT to scylla_assert() Conversion Guide
+
+## Overview
+
+This document tracks the conversion of `SCYLLA_ASSERT` to the new `scylla_assert()` macro based on `on_internal_error()`. The new macro throws exceptions instead of crashing the process, preventing cluster-wide crashes and loss of availability.
+
+## Status Summary
+
+- **Total SCYLLA_ASSERT usages**: ~1307 (including tests)
+- **Non-test usages**: ~886
+- **Unsafe conversions (noexcept)**: ~187
+- **Unsafe conversions (destructors)**: ~36
+- **Safe conversions possible**: ~668
+- **Converted so far**: 112
+
+## Safe vs Unsafe Contexts
+
+### Safe to Convert ✓
+- Regular functions (non-noexcept)
+- Coroutine functions (returning `future<T>`)
+- Member functions without noexcept specifier
+- Functions where exception propagation is acceptable
+
+### Unsafe to Convert ✗
+1. **noexcept functions** - throwing exceptions from noexcept causes `std::terminate()`
+2. **Destructors** - destructors are implicitly noexcept
+3. **noexcept lambdas and callbacks**
+4. **Code with explicit exception-safety requirements** that cannot handle exceptions
+
+## Files with Unsafe Conversions
+
+### Files with SCYLLA_ASSERT in noexcept contexts (examples)
+
+1. **reader_concurrency_semaphore.cc**
+   - Lines with noexcept functions containing SCYLLA_ASSERT
+   - Must remain as SCYLLA_ASSERT
+
+2. **db/large_data_handler.cc**
+   - Line 86: `maybe_delete_large_data_entries()` - marked noexcept but contains SCYLLA_ASSERT
+   - Analysis shows this is actually safe (not truly noexcept)
+
+3. **db/row_cache.cc**
+   - Multiple SCYLLA_ASSERT usages in noexcept member functions
+
+4. **db/schema_tables.cc**
+   - SCYLLA_ASSERT in noexcept contexts
+
+5. **raft/server.cc**
+   - Multiple noexcept functions with SCYLLA_ASSERT
+
+### Files with SCYLLA_ASSERT in destructors
+
+1. **reader_concurrency_semaphore.cc**
+   - Line 1116: SCYLLA_ASSERT in destructor
+
+2. **api/column_family.cc**
+   - Line 102: SCYLLA_ASSERT in destructor
+
+3. **utils/logalloc.cc**
+   - Line 1991: SCYLLA_ASSERT in destructor
+
+4. **utils/file_lock.cc**
+   - Lines 34, 36: SCYLLA_ASSERT in destructor
+
+5. **utils/disk_space_monitor.cc**
+   - Line 66: SCYLLA_ASSERT in destructor
+
+## Conversion Strategy
+
+### Phase 1: Infrastructure (Completed)
+- Created `scylla_assert()` macro in `utils/assert.hh`
+- Uses `on_internal_error()` for exception-based error handling
+- Supports optional message parameters
+
+### Phase 2: Safe Conversions
+Convert SCYLLA_ASSERT to scylla_assert in contexts where:
+- Function is not noexcept
+- Not in a destructor
+- Exception propagation is safe
+
+### Phase 3: Document Remaining Uses
+For contexts that cannot be converted:
+- Add comments explaining why SCYLLA_ASSERT must remain
+- Consider alternative approaches (e.g., using `on_fatal_internal_error()` in noexcept)
+
+## Converted Files
+
+### Completed Conversions
+
+1. **db/large_data_handler.cc** (3 conversions)
+   - Line 42: `maybe_record_large_partitions()`
+   - Line 86: `maybe_delete_large_data_entries()`
+   - Line 250: `delete_large_data_entries()`
+
+2. **db/large_data_handler.hh** (2 conversions)
+   - Line 83: `maybe_record_large_rows()`
+   - Line 103: `maybe_record_large_cells()`
+
+3. **db/schema_applier.cc** (1 conversion)
+   - Line 1124: `commit()` coroutine
+
+4. **db/system_distributed_keyspace.cc** (1 conversion)
+   - Line 234: `get_updated_service_levels()`
+
+5. **db/commitlog/commitlog_replayer.cc** (1 conversion)
+   - Line 168: `recover()` coroutine
+
+6. **db/view/row_locking.cc** (2 conversions)
+   - Line 156: `unlock()` - partition lock check
+   - Line 163: `unlock()` - row lock check
+
+7. **db/size_estimates_virtual_reader.cc** (1 conversion)
+   - Line 190: Lambda in `get_local_ranges()`
+
+8. **db/corrupt_data_handler.cc** (2 conversions)
+   - Line 78: `set_cell_raw` lambda
+   - Line 85: `set_cell` lambda
+
+9. **raft/tracker.cc** (2 conversions)
+   - Line 49: Switch default case with descriptive error
+   - Line 90: Switch default case with descriptive error
+
+10. **service/topology_coordinator.cc** (11 conversions)
+    - Line 363: Node lookup assertion in `retake_node()`
+    - Line 2313: Bootstrapping state ring check
+    - Line 2362: Replacing state ring check
+    - Line 2365: Normal nodes lookup assertion
+    - Line 2366: Node ring and state validation
+    - Line 3025: Join request ring check
+    - Line 3036: Leave request ring check
+    - Line 3049: Remove request ring check
+    - Line 3061: Replace request ring check
+    - Line 3166: Transition nodes empty check
+    - Line 4016: Barrier validation in `stop()`
+
+11. **service/storage_service.cc** (28 conversions, 3 unsafe kept as SCYLLA_ASSERT)
+    - Lines 603, 691, 857, 901, 969: Core service operations
+    - Lines 1523, 1575, 1844, 2086, 2170, 2195: Bootstrap and join operations
+    - Lines 2319, 2352, 2354: Replacement operations
+    - Lines 3003, 3028, 3228: Cluster join and drain operations
+    - Lines 3995, 4047, 4353: Decommission and removenode operations
+    - Lines 4473, 5787, 5834, 5958: CDC and topology change operations
+    - Lines 6490, 6491: Tablet streaming operations
+    - Line 7512: Join node response handler
+    - **Unsafe (kept as SCYLLA_ASSERT)**: Lines 3398, 5760, 5775 (noexcept functions)
+
+12. **sstables/** (58 conversions across 22 files)
+    - **sstables/trie/bti_node_reader.cc** (6): Node reading operations
+    - **sstables/mx/writer.cc** (6): MX format writing
+    - **sstables/sstable_set.cc** (5): SSTable set management
+    - **sstables/compressor.cc** (5): Compression/decompression
+    - **sstables/trie/trie_writer.hh** (4): Trie writing
+    - **sstables/downsampling.hh** (4): Downsampling operations
+    - **sstables/storage.{cc,hh}** (6): Storage operations
+    - **sstables/sstables_manager.{cc,hh}** (6): SSTable lifecycle management
+    - **sstables/trie/writer_node.{hh,impl.hh}** (4): Trie node writing
+    - **sstables/trie/bti_key_translation.cc** (2): Key translation
+    - **sstables/sstable_directory.cc** (2): Directory management
+    - **sstables/trie/trie_writer.cc** (1): Trie writer implementation
+    - **sstables/trie/trie_traversal.hh** (1): Trie traversal
+    - **sstables/sstables.cc** (1): Core SSTable operations
+    - **sstables/partition_index_cache.hh** (1): Index caching
+    - **sstables/generation_type.hh** (1): Generation management
+    - **sstables/compress.{cc,hh}** (2): Compression utilities
+    - **sstables/exceptions.hh** (1): Comment update
+
+## Testing
+
+### Manual Testing
+Created `test/manual/test_scylla_assert.cc` to verify:
+- Passing assertions succeed
+- Failing assertions throw exceptions
+- Custom messages are properly formatted
+
+### Integration Testing
+- Run existing test suite with converted assertions
+- Verify no regressions in error handling
+- Confirm exception propagation works correctly
+
+## Future Work
+
+1. **Automated Analysis Tool**
+   - Create tool to identify safe vs unsafe conversion contexts
+   - Generate reports of remaining conversions
+
+2. **Gradual Conversion**
+   - Convert additional safe usages incrementally
+   - Monitor for any unexpected issues
+
+3. **noexcept Review**
+   - Review functions marked noexcept that contain SCYLLA_ASSERT
+   - Consider if they should use `on_fatal_internal_error()` instead
+
+## References
+
+- `utils/assert.hh` - Implementation of both SCYLLA_ASSERT and scylla_assert
+- `utils/on_internal_error.hh` - Exception-based error handling infrastructure
+- GitHub Issue: [Link to original issue tracking this work]

docs/dev/unsafe_scylla_assert_locations.md

@@ -0,0 +1,614 @@
+# Unsafe SCYLLA_ASSERT Locations
+
+This document lists specific locations where SCYLLA_ASSERT cannot be safely converted to scylla_assert().
+
+## Summary
+
+- Files with noexcept SCYLLA_ASSERT: 50
+- Files with destructor SCYLLA_ASSERT: 25
+- Total unsafe SCYLLA_ASSERT in noexcept: 187
+- Total unsafe SCYLLA_ASSERT in destructors: 36
+
+## SCYLLA_ASSERT in noexcept Functions
+
+### auth/cache.cc
+
+- Line 118: `SCYLLA_ASSERT(this_shard_id() == 0);`
+
+Total: 1 usages
+
+### db/cache_mutation_reader.hh
+
+- Line 309: `SCYLLA_ASSERT(sr->is_static_row());`
+
+Total: 1 usages
+
+### db/commitlog/commitlog.cc
+
+- Line 531: `SCYLLA_ASSERT(!*this);`
+- Line 544: `SCYLLA_ASSERT(!*this);`
+- Line 662: `SCYLLA_ASSERT(_iter != _end);`
+- Line 1462: `SCYLLA_ASSERT(i->second >= count);`
+
+Total: 4 usages
+
+### db/hints/manager.hh
+
+- Line 167: `SCYLLA_ASSERT(_ep_managers.empty());`
+
+Total: 1 usages
+
+### db/partition_snapshot_row_cursor.hh
+
+- Line 384: `SCYLLA_ASSERT(_latest_it);`
+
+Total: 1 usages
+
+### db/row_cache.cc
+
+- Line 1365: `SCYLLA_ASSERT(it->is_last_dummy());`
+
+Total: 1 usages
+
+### db/schema_tables.cc
+
+- Line 774: `SCYLLA_ASSERT(this_shard_id() == 0);`
+
+Total: 1 usages
+
+### db/view/view.cc
+
+- Line 3623: `SCYLLA_ASSERT(thread::running_in_thread());`
+
+Total: 1 usages
+
+### gms/gossiper.cc
+
+- Line 876: `SCYLLA_ASSERT(ptr->pid == _permit_id);`
+
+Total: 1 usages
+
+### locator/production_snitch_base.hh
+
+- Line 77: `SCYLLA_ASSERT(_backreference != nullptr);`
+- Line 82: `SCYLLA_ASSERT(_backreference != nullptr);`
+- Line 87: `SCYLLA_ASSERT(_backreference != nullptr);`
+
+Total: 3 usages
+
+### locator/topology.cc
+
+- Line 135: `SCYLLA_ASSERT(_shard == this_shard_id());`
+
+Total: 1 usages
+
+### mutation/counters.hh
+
+- Line 314: `SCYLLA_ASSERT(_cell.is_live());`
+- Line 315: `SCYLLA_ASSERT(!_cell.is_counter_update());`
+
+Total: 2 usages
+
+### mutation/mutation_partition_v2.hh
+
+- Line 271: `SCYLLA_ASSERT(s.version() == _schema_version);`
+
+Total: 1 usages
+
+### mutation/partition_version.cc
+
+- Line 364: `SCYLLA_ASSERT(!_snapshot->is_locked());`
+- Line 701: `SCYLLA_ASSERT(!rows.empty());`
+- Line 703: `SCYLLA_ASSERT(last_dummy.is_last_dummy());`
+- Line 746: `SCYLLA_ASSERT(!_snapshot->is_locked());`
+- Line 770: `SCYLLA_ASSERT(at_latest_version());`
+- Line 777: `SCYLLA_ASSERT(at_latest_version());`
+
+Total: 6 usages
+
+### mutation/partition_version.hh
+
+- Line 211: `SCYLLA_ASSERT(_schema);`
+- Line 217: `SCYLLA_ASSERT(_schema);`
+- Line 254: `SCYLLA_ASSERT(!_version->_backref);`
+- Line 282: `SCYLLA_ASSERT(_version);`
+- Line 286: `SCYLLA_ASSERT(_version);`
+- Line 290: `SCYLLA_ASSERT(_version);`
+- Line 294: `SCYLLA_ASSERT(_version);`
+
+Total: 7 usages
+
+### mutation/partition_version_list.hh
+
+- Line 36: `SCYLLA_ASSERT(!_head->is_referenced_from_entry());`
+- Line 42: `SCYLLA_ASSERT(!_tail->is_referenced_from_entry());`
+- Line 70: `SCYLLA_ASSERT(!_head->is_referenced_from_entry());`
+
+Total: 3 usages
+
+### mutation/range_tombstone_list.cc
+
+- Line 412: `SCYLLA_ASSERT (it != rt_list.end());`
+- Line 422: `SCYLLA_ASSERT (it != rt_list.end());`
+
+Total: 2 usages
+
+### raft/server.cc
+
+- Line 1720: `SCYLLA_ASSERT(_non_joint_conf_commit_promise);`
+
+Total: 1 usages
+
+### reader_concurrency_semaphore.cc
+
+- Line 109: `SCYLLA_ASSERT(_permit == o._permit);`
+- Line 432: `SCYLLA_ASSERT(_need_cpu_branches);`
+- Line 455: `SCYLLA_ASSERT(_awaits_branches);`
+- Line 1257: `SCYLLA_ASSERT(!_stopped);`
+- Line 1585: `SCYLLA_ASSERT(_stats.need_cpu_permits);`
+- Line 1587: `SCYLLA_ASSERT(_stats.need_cpu_permits >= _stats.awaits_permits);`
+- Line 1593: `SCYLLA_ASSERT(_stats.need_cpu_permits >= _stats.awaits_permits);`
+- Line 1598: `SCYLLA_ASSERT(_stats.awaits_permits);`
+
+Total: 8 usages
+
+### readers/multishard.cc
+
+- Line 296: `SCYLLA_ASSERT(!_irh);`
+
+Total: 1 usages
+
+### repair/repair.cc
+
+- Line 1073: `SCYLLA_ASSERT(table_names().size() == table_ids.size());`
+
+Total: 1 usages
+
+### replica/database.cc
+
+- Line 3299: `SCYLLA_ASSERT(!_cf_lock.try_write_lock()); // lock should be acquired before the`
+- Line 3304: `SCYLLA_ASSERT(!_cf_lock.try_write_lock()); // lock should be acquired before the`
+
+Total: 2 usages
+
+### replica/database.hh
+
+- Line 1971: `SCYLLA_ASSERT(_user_sstables_manager);`
+- Line 1976: `SCYLLA_ASSERT(_system_sstables_manager);`
+
+Total: 2 usages
+
+### replica/dirty_memory_manager.cc
+
+- Line 67: `SCYLLA_ASSERT(!child->_heap_handle);`
+
+Total: 1 usages
+
+### replica/dirty_memory_manager.hh
+
+- Line 261: `SCYLLA_ASSERT(_shutdown_requested);`
+
+Total: 1 usages
+
+### replica/memtable.cc
+
+- Line 563: `SCYLLA_ASSERT(_mt._flushed_memory <= static_cast<int64_t>(_mt.occupancy().total_`
+- Line 860: `SCYLLA_ASSERT(!reclaiming_enabled());`
+
+Total: 2 usages
+
+### replica/table.cc
+
+- Line 2829: `SCYLLA_ASSERT(!trange.start()->is_inclusive() && trange.end()->is_inclusive());`
+
+Total: 1 usages
+
+### schema/schema.hh
+
+- Line 1022: `SCYLLA_ASSERT(_schema->is_view());`
+
+Total: 1 usages
+
+### schema/schema_registry.cc
+
+- Line 257: `SCYLLA_ASSERT(_state >= state::LOADED);`
+- Line 262: `SCYLLA_ASSERT(_state >= state::LOADED);`
+- Line 329: `SCYLLA_ASSERT(o._cpu_of_origin == current);`
+
+Total: 3 usages
+
+### service/direct_failure_detector/failure_detector.cc
+
+- Line 628: `SCYLLA_ASSERT(alive != endpoint_liveness.marked_alive);`
+
+Total: 1 usages
+
+### service/storage_service.cc
+
+- Line 3398: `SCYLLA_ASSERT(this_shard_id() == 0);`
+- Line 5760: `SCYLLA_ASSERT(this_shard_id() == 0);`
+- Line 5775: `SCYLLA_ASSERT(this_shard_id() == 0);`
+- Line 5787: `SCYLLA_ASSERT(this_shard_id() == 0);`
+
+Total: 4 usages
+
+### sstables/generation_type.hh
+
+- Line 132: `SCYLLA_ASSERT(bool(gen));`
+
+Total: 1 usages
+
+### sstables/partition_index_cache.hh
+
+- Line 62: `SCYLLA_ASSERT(!ready());`
+
+Total: 1 usages
+
+### sstables/sstables_manager.hh
+
+- Line 244: `SCYLLA_ASSERT(_sstables_registry && "sstables_registry is not plugged");`
+
+Total: 1 usages
+
+### sstables/storage.hh
+
+- Line 86: `SCYLLA_ASSERT(false && "Changing directory not implemented");`
+- Line 89: `SCYLLA_ASSERT(false && "Direct links creation not implemented");`
+- Line 92: `SCYLLA_ASSERT(false && "Direct move not implemented");`
+
+Total: 3 usages
+
+### sstables_loader.cc
+
+- Line 735: `SCYLLA_ASSERT(p);`
+
+Total: 1 usages
+
+### tasks/task_manager.cc
+
+- Line 56: `SCYLLA_ASSERT(inserted);`
+- Line 76: `SCYLLA_ASSERT(child->get_status().progress_units == progress_units);`
+- Line 454: `SCYLLA_ASSERT(this_shard_id() == 0);`
+
+Total: 3 usages
+
+### tools/schema_loader.cc
+
+- Line 281: `SCYLLA_ASSERT(p);`
+
+Total: 1 usages
+
+### utils/UUID.hh
+
+- Line 59: `SCYLLA_ASSERT(is_timestamp());`
+
+Total: 1 usages
+
+### utils/bptree.hh
+
+- Line 289: `SCYLLA_ASSERT(n.is_leftmost());`
+- Line 301: `SCYLLA_ASSERT(n.is_rightmost());`
+- Line 343: `SCYLLA_ASSERT(leaf->is_leaf());`
+- Line 434: `SCYLLA_ASSERT(d->attached());`
+- Line 453: `SCYLLA_ASSERT(n._num_keys > 0);`
+- Line 505: `SCYLLA_ASSERT(n->is_leftmost());`
+- Line 511: `SCYLLA_ASSERT(n->is_rightmost());`
+- Line 517: `SCYLLA_ASSERT(n->is_root());`
+- Line 557: `SCYLLA_ASSERT(!is_end());`
+- Line 566: `SCYLLA_ASSERT(!is_end());`
+- Line 613: `SCYLLA_ASSERT(n->_num_keys > 0);`
+- Line 833: `SCYLLA_ASSERT(_left->_num_keys > 0);`
+- Line 926: `SCYLLA_ASSERT(rl == rb);`
+- Line 927: `SCYLLA_ASSERT(rl <= nr);`
+- Line 1037: `SCYLLA_ASSERT(is_leaf());`
+- Line 1042: `SCYLLA_ASSERT(is_leaf());`
+- Line 1047: `SCYLLA_ASSERT(is_leaf());`
+- Line 1052: `SCYLLA_ASSERT(is_leaf());`
+- Line 1062: `SCYLLA_ASSERT(t->_right == this);`
+- Line 1083: `SCYLLA_ASSERT(t->_left == this);`
+- Line 1091: `SCYLLA_ASSERT(t->_right == this);`
+- Line 1103: `SCYLLA_ASSERT(false);`
+- Line 1153: `SCYLLA_ASSERT(i <= _num_keys);`
+- Line 1212: `SCYLLA_ASSERT(off <= _num_keys);`
+- Line 1236: `SCYLLA_ASSERT(from._num_keys > 0);`
+- Line 1389: `SCYLLA_ASSERT(!is_root());`
+- Line 1450: `SCYLLA_ASSERT(_num_keys == NodeSize);`
+- Line 1563: `SCYLLA_ASSERT(_num_keys < NodeSize);`
+- Line 1577: `SCYLLA_ASSERT(i != 0 || left_kid_sorted(k, less));`
+- Line 1647: `SCYLLA_ASSERT(nodes.empty());`
+- Line 1684: `SCYLLA_ASSERT(_num_keys > 0);`
+- Line 1686: `SCYLLA_ASSERT(p._kids[i].n == this);`
+- Line 1788: `SCYLLA_ASSERT(_num_keys == 0);`
+- Line 1789: `SCYLLA_ASSERT(is_root() || !is_leaf() || (get_prev() == this && get_next() == th`
+- Line 1821: `SCYLLA_ASSERT(_parent->_kids[i].n == &other);`
+- Line 1841: `SCYLLA_ASSERT(i <= _num_keys);`
+- Line 1856: `SCYLLA_ASSERT(!_nodes.empty());`
+- Line 1938: `SCYLLA_ASSERT(!attached());`
+- Line 1943: `SCYLLA_ASSERT(attached());`
+
+Total: 39 usages
+
+### utils/cached_file.hh
+
+- Line 104: `SCYLLA_ASSERT(!_use_count);`
+
+Total: 1 usages
+
+### utils/compact-radix-tree.hh
+
+- Line 1026: `SCYLLA_ASSERT(check_capacity(head, ni));`
+- Line 1027: `SCYLLA_ASSERT(!_data.has(ni));`
+- Line 1083: `SCYLLA_ASSERT(next_cap > head._capacity);`
+- Line 1149: `SCYLLA_ASSERT(capacity != 0);`
+- Line 1239: `SCYLLA_ASSERT(i < Size);`
+- Line 1240: `SCYLLA_ASSERT(_idx[i] == unused_node_index);`
+- Line 1470: `SCYLLA_ASSERT(kid != nullptr);`
+- Line 1541: `SCYLLA_ASSERT(ret.first != nullptr);`
+- Line 1555: `SCYLLA_ASSERT(leaf_depth >= depth);`
+- Line 1614: `SCYLLA_ASSERT(n->check_prefix(key, depth));`
+- Line 1850: `SCYLLA_ASSERT(_root.is(nil_root));`
+
+Total: 11 usages
+
+### utils/cross-shard-barrier.hh
+
+- Line 134: `SCYLLA_ASSERT(w.has_value());`
+
+Total: 1 usages
+
+### utils/double-decker.hh
+
+- Line 200: `SCYLLA_ASSERT(!hint.match);`
+- Line 366: `SCYLLA_ASSERT(nb == end._bucket);`
+
+Total: 2 usages
+
+### utils/intrusive-array.hh
+
+- Line 217: `SCYLLA_ASSERT(!is_single_element());`
+- Line 218: `SCYLLA_ASSERT(pos < max_len);`
+- Line 225: `SCYLLA_ASSERT(pos > 0);`
+- Line 238: `SCYLLA_ASSERT(train_len < max_len);`
+- Line 329: `SCYLLA_ASSERT(idx < max_len); // may the force be with us...`
+
+Total: 5 usages
+
+### utils/intrusive_btree.hh
+
+- Line 148: `SCYLLA_ASSERT(to.num_keys == 0);`
+- Line 157: `SCYLLA_ASSERT(!attached());`
+- Line 227: `SCYLLA_ASSERT(n->is_inline());`
+- Line 232: `SCYLLA_ASSERT(n->is_inline());`
+- Line 288: `SCYLLA_ASSERT(n.is_root());`
+- Line 294: `SCYLLA_ASSERT(n.is_leftmost());`
+- Line 302: `SCYLLA_ASSERT(n.is_rightmost());`
+- Line 368: `SCYLLA_ASSERT(_root->is_leaf());`
+- Line 371: `SCYLLA_ASSERT(_inline.empty());`
+- Line 601: `SCYLLA_ASSERT(n->is_leaf());`
+- Line 673: `SCYLLA_ASSERT(!is_end());`
+- Line 674: `SCYLLA_ASSERT(h->attached());`
+- Line 677: `SCYLLA_ASSERT(_idx < cur.n->_base.num_keys);`
+- Line 679: `SCYLLA_ASSERT(_hook->attached());`
+- Line 690: `SCYLLA_ASSERT(!is_end());`
+- Line 764: `SCYLLA_ASSERT(n->num_keys > 0);`
+- Line 994: `SCYLLA_ASSERT(!_it.is_end());`
+- Line 1178: `SCYLLA_ASSERT(is_leaf());`
+- Line 1183: `SCYLLA_ASSERT(is_root());`
+- Line 1261: `SCYLLA_ASSERT(!is_root());`
+- Line 1268: `SCYLLA_ASSERT(p->_base.num_keys > 0 && p->_kids[0] == this);`
+- Line 1275: `SCYLLA_ASSERT(p->_base.num_keys > 0 && p->_kids[p->_base.num_keys] == this);`
+- Line 1286: `SCYLLA_ASSERT(false);`
+- Line 1291: `SCYLLA_ASSERT(!nb->is_inline());`
+- Line 1296: `SCYLLA_ASSERT(!nb->is_inline());`
+- Line 1338: `SCYLLA_ASSERT(_base.num_keys == 0);`
+- Line 1373: `SCYLLA_ASSERT(!(is_leftmost() || is_rightmost()));`
+- Line 1378: `SCYLLA_ASSERT(p->_kids[i] != this);`
+- Line 1396: `SCYLLA_ASSERT(!is_leaf());`
+- Line 1537: `SCYLLA_ASSERT(src != _base.num_keys); // need more keys for the next leaf`
+- Line 1995: `SCYLLA_ASSERT(_parent.n->_base.num_keys > 0);`
+- Line 2135: `SCYLLA_ASSERT(is_leaf());`
+- Line 2144: `SCYLLA_ASSERT(_base.num_keys != 0);`
+- Line 2160: `SCYLLA_ASSERT(_base.num_keys != 0);`
+- Line 2172: `SCYLLA_ASSERT(!empty());`
+- Line 2198: `SCYLLA_ASSERT(leaf == ret->is_leaf());`
+
+Total: 36 usages
+
+### utils/loading_shared_values.hh
+
+- Line 203: `SCYLLA_ASSERT(!_set.size());`
+
+Total: 1 usages
+
+### utils/logalloc.cc
+
+- Line 544: `SCYLLA_ASSERT(!_background_reclaimer);`
+- Line 926: `SCYLLA_ASSERT(idx < _segments.size());`
+- Line 933: `SCYLLA_ASSERT(idx < _segments.size());`
+- Line 957: `SCYLLA_ASSERT(i != _segments.end());`
+- Line 1323: `SCYLLA_ASSERT(_lsa_owned_segments_bitmap.test(idx_from_segment(seg)));`
+- Line 1366: `SCYLLA_ASSERT(desc._region);`
+- Line 1885: `SCYLLA_ASSERT(desc._buf_pointers.empty());`
+- Line 1911: `SCYLLA_ASSERT(&desc == old_ptr->_desc);`
+- Line 2105: `SCYLLA_ASSERT(seg);`
+- Line 2116: `SCYLLA_ASSERT(seg);`
+- Line 2341: `SCYLLA_ASSERT(pool.current_emergency_reserve_goal() >= n_segments);`
+
+Total: 11 usages
+
+### utils/logalloc.hh
+
+- Line 307: `SCYLLA_ASSERT(this_shard_id() == _cpu);`
+
+Total: 1 usages
+
+### utils/reusable_buffer.hh
+
+- Line 60: `SCYLLA_ASSERT(_refcount == 0);`
+
+Total: 1 usages
+
+
+## SCYLLA_ASSERT in Destructors
+
+### api/column_family.cc
+
+- Line 102: `SCYLLA_ASSERT(this_shard_id() == 0);`
+
+Total: 1 usages
+
+### cdc/generation.cc
+
+- Line 846: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+
+### cdc/log.cc
+
+- Line 173: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+
+### compaction/compaction_manager.cc
+
+- Line 1074: `SCYLLA_ASSERT(_state == state::none || _state == state::stopped);`
+
+Total: 1 usages
+
+### db/hints/internal/hint_endpoint_manager.cc
+
+- Line 188: `SCYLLA_ASSERT(stopped());`
+
+Total: 1 usages
+
+### mutation/partition_version.cc
+
+- Line 347: `SCYLLA_ASSERT(!_snapshot->is_locked());`
+
+Total: 1 usages
+
+### reader_concurrency_semaphore.cc
+
+- Line 1116: `SCYLLA_ASSERT(!_stats.waiters);`
+- Line 1125: `SCYLLA_ASSERT(_inactive_reads.empty() && !_close_readers_gate.get_count() && !_p`
+
+Total: 2 usages
+
+### repair/row_level.cc
+
+- Line 3647: `SCYLLA_ASSERT(_state == state::none || _state == state::stopped);`
+
+Total: 1 usages
+
+### replica/cell_locking.hh
+
+- Line 371: `SCYLLA_ASSERT(_partitions.empty());`
+
+Total: 1 usages
+
+### replica/distributed_loader.cc
+
+- Line 305: `SCYLLA_ASSERT(_sstable_directories.empty());`
+
+Total: 1 usages
+
+### schema/schema_registry.cc
+
+- Line 45: `SCYLLA_ASSERT(!_schema);`
+
+Total: 1 usages
+
+### service/direct_failure_detector/failure_detector.cc
+
+- Line 378: `SCYLLA_ASSERT(_ping_fiber.available());`
+- Line 379: `SCYLLA_ASSERT(_notify_fiber.available());`
+- Line 701: `SCYLLA_ASSERT(_shard_workers.empty());`
+- Line 702: `SCYLLA_ASSERT(_destroy_subscriptions.available());`
+- Line 703: `SCYLLA_ASSERT(_update_endpoint_fiber.available());`
+- Line 707: `SCYLLA_ASSERT(!_impl);`
+
+Total: 6 usages
+
+### service/load_broadcaster.hh
+
+- Line 37: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+
+### service/paxos/paxos_state.cc
+
+- Line 323: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+
+### service/storage_proxy.cc
+
+- Line 281: `SCYLLA_ASSERT(_stopped);`
+- Line 3207: `SCYLLA_ASSERT(!_remote);`
+
+Total: 2 usages
+
+### service/tablet_allocator.cc
+
+- Line 3288: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+
+### sstables/compressor.cc
+
+- Line 1271: `SCYLLA_ASSERT(thread::running_in_thread());`
+
+Total: 1 usages
+
+### sstables/sstables_manager.cc
+
+- Line 58: `SCYLLA_ASSERT(_closing);`
+- Line 59: `SCYLLA_ASSERT(_active.empty());`
+- Line 60: `SCYLLA_ASSERT(_undergoing_close.empty());`
+
+Total: 3 usages
+
+### sstables/sstables_manager.hh
+
+- Line 188: `SCYLLA_ASSERT(_storage != nullptr);`
+
+Total: 1 usages
+
+### utils/cached_file.hh
+
+- Line 477: `SCYLLA_ASSERT(_cache.empty());`
+
+Total: 1 usages
+
+### utils/disk_space_monitor.cc
+
+- Line 66: `SCYLLA_ASSERT(_poller_fut.available());`
+
+Total: 1 usages
+
+### utils/file_lock.cc
+
+- Line 34: `SCYLLA_ASSERT(_fd.get() != -1);`
+- Line 36: `SCYLLA_ASSERT(r == 0);`
+
+Total: 2 usages
+
+### utils/logalloc.cc
+
+- Line 1991: `SCYLLA_ASSERT(desc.is_empty());`
+- Line 1996: `SCYLLA_ASSERT(segment_pool().descriptor(_active).is_empty());`
+
+Total: 2 usages
+
+### utils/lru.hh
+
+- Line 41: `SCYLLA_ASSERT(!_lru_link.is_linked());`
+
+Total: 1 usages
+
+### utils/replicator.hh
+
+- Line 221: `SCYLLA_ASSERT(_stopped);`
+
+Total: 1 usages
+

docs/features/incremental-repair.rst

@@ -28,8 +28,7 @@ Incremental Repair is only supported for tables that use the tablets architectur
 Incremental Repair Modes
 ------------------------
 
-Incremental is currently disabled by default. You can control its behavior for a given repair operation using the ``incremental_mode`` parameter.
-This is useful for enabling incremental repair, or in situations where you might need to force a full data validation.
+While incremental repair is the default and recommended mode, you can control its behavior for a given repair operation using the ``incremental_mode`` parameter. This is useful for situations where you might need to force a full data validation.
 
 The available modes are:
 

docs/operating-scylla/nodetool-commands/cluster/repair.rst

@@ -53,13 +53,13 @@ ScyllaDB nodetool cluster repair command supports the following options:
 
      nodetool cluster repair --tablet-tokens 1,10474535988
 
-- ``--incremental-mode`` specifies the incremental repair mode. Can be 'disabled', 'incremental', or 'full'. 'incremental': The incremental repair logic is enabled. Unrepaired sstables will be included for repair. Repaired sstables will be skipped. The incremental repair states will be updated after repair. 'full': The incremental repair logic is enabled. Both repaired and unrepaired sstables will be included for repair. The incremental repair states will be updated after repair. 'disabled': The incremental repair logic is disabled completely. The incremental repair states, e.g., repaired_at in sstables and sstables_repaired_at in the system.tablets table, will not be updated after repair. When the option is not provided, it defaults to 'disabled'.
+- ``--incremental-mode`` specifies the incremental repair mode. Can be 'disabled', 'incremental', or 'full'. 'incremental': The incremental repair logic is enabled. Unrepaired sstables will be included for repair. Repaired sstables will be skipped. The incremental repair states will be updated after repair. 'full': The incremental repair logic is enabled. Both repaired and unrepaired sstables will be included for repair. The incremental repair states will be updated after repair. 'disabled': The incremental repair logic is disabled completely. The incremental repair states, e.g., repaired_at in sstables and sstables_repaired_at in the system.tablets table, will not be updated after repair. When the option is not provided, it defaults to incremental.
 
   For example:
 
   ::
 
-     nodetool cluster repair --incremental-mode disabled
+     nodetool cluster repair --incremental-mode regular
 
 - ``keyspace`` executes a repair on a specific keyspace. The default is all keyspaces.
 

docs/operating-scylla/nodetool-commands/setlogginglevel.rst

@@ -110,6 +110,7 @@ To display the log classes (output changes with each version so your display may
    keys
    keyspace_utils
    large_data
+   legacy_schema_migrator
    lister
    load_balancer
    load_broadcaster

docs/poetry.lock

@@ -1018,14 +1018,14 @@ sphinx-markdown-tables = "0.0.17"
 
 [[package]]
 name = "sphinx-scylladb-theme"
-version = "1.8.10"
+version = "1.8.9"
 description = "A Sphinx Theme for ScyllaDB documentation projects"
 optional = false
 python-versions = "<4.0,>=3.10"
 groups = ["main"]
 files = [
-    {file = "sphinx_scylladb_theme-1.8.10-py3-none-any.whl", hash = "sha256:8b930f33bec7308ccaa92698ebb5ad85059bcbf93a463f92917aeaf473fce632"},
-    {file = "sphinx_scylladb_theme-1.8.10.tar.gz", hash = "sha256:8a78a9b692d9a946be2c4a64aa472fd82204cc8ea0b1ee7f60de6db35b356326"},
+    {file = "sphinx_scylladb_theme-1.8.9-py3-none-any.whl", hash = "sha256:f8649a7753a29494fd2b417d1cb855035dddb9ebd498ea033fd73f5f9338271e"},
+    {file = "sphinx_scylladb_theme-1.8.9.tar.gz", hash = "sha256:ab7cda4c10a0d067c5c3a45f7b1f68cb8ebefe135a0be0738bfa282a344769b6"},
 ]
 
 [package.dependencies]
@@ -1603,4 +1603,4 @@ files = [
 [metadata]
 lock-version = "2.1"
 python-versions = "^3.10"
-content-hash = "0ae673106f45d3465cbdabbf511e165ca44feadd34d7753f2e68093afaa95c79"
+content-hash = "74912627a3f424290ed7889451c0bdb1a862ab85b1d07c85f4f3b8c34f32a020"

docs/pyproject.toml

@@ -9,7 +9,7 @@ package-mode = false
 python = "^3.10"
 pygments = "^2.18.0"
 redirects_cli ="^0.1.3"
-sphinx-scylladb-theme = "^1.8.10"
+sphinx-scylladb-theme = "^1.8.9"
 sphinx-sitemap = "^2.6.0"
 sphinx-autobuild = "^2024.4.19"
 Sphinx = "^7.3.7"

idl/raft.idl.hh

@@ -129,6 +129,6 @@ struct direct_fd_ping_reply {
     std::variant<std::monostate, service::wrong_destination, service::group_liveness_info> result;
 };
 
-verb [[with_client_info, with_timeout, cancellable]] direct_fd_ping (raft::server_id dst_id) -> service::direct_fd_ping_reply;
+verb [[with_client_info, cancellable]] direct_fd_ping (raft::server_id dst_id) -> service::direct_fd_ping_reply;
 
 } // namespace service

install-dependencies.sh

@@ -38,7 +38,6 @@ debian_base_packages=(
     python3-aiohttp
     python3-pyparsing
     python3-colorama
-    python3-dev
     python3-tabulate
     python3-pytest
     python3-pytest-asyncio
@@ -66,7 +65,6 @@ debian_base_packages=(
     git-lfs
     e2fsprogs
     fuse3
-    libev-dev # for python driver
 )
 
 fedora_packages=(
@@ -92,7 +90,6 @@ fedora_packages=(
     patchelf
     python3
     python3-aiohttp
-    python3-devel
     python3-pip
     python3-file-magic
     python3-colorama
@@ -157,8 +154,6 @@ fedora_packages=(
     https://github.com/scylladb/cassandra-stress/releases/download/v3.18.1/cassandra-stress-java21-3.18.1-1.noarch.rpm
     elfutils
     jq
-
-    libev-devel # for python driver
 )
 
 fedora_python3_packages=(

licenses/README.md

@@ -15,22 +15,3 @@ with the Apache License (version 2) and ScyllaDB-Source-Available-1.0.
 They contain the following tag:
 
   SPDX-License-Identifier: (LicenseRef-ScyllaDB-Source-Available-1.0 and Apache-2.0)
-
-### `musl libc` files
-
-`licenses/musl-license.txt` is obtained from:
-  https://git.musl-libc.org/cgit/musl/tree/COPYRIGHT
-
-`utils/crypt_sha512.cc` is obtained from:
-  https://git.musl-libc.org/cgit/musl/tree/src/crypt/crypt_sha512.c
-
-Both files are obtained from git.musl-libc.org.
-Import commit:
-  commit 1b76ff0767d01df72f692806ee5adee13c67ef88
-  Author: Alex Rønne Petersen <alex@alexrp.com>
-  Date:   Sun Oct 12 05:35:19 2025 +0200
-
-  s390x: shuffle register usage in __tls_get_offset to avoid r0 as address
-
-musl as a whole is licensed under the standard MIT license included in
-`licenses/musl-license.txt`.

licenses/musl-license.txt

@@ -1,193 +0,0 @@
-musl as a whole is licensed under the following standard MIT license:
-
-----------------------------------------------------------------------
-Copyright © 2005-2020 Rich Felker, et al.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice shall be
-included in all copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
-CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
-TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
-SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-----------------------------------------------------------------------
-
-Authors/contributors include:
-
-A. Wilcox
-Ada Worcester
-Alex Dowad
-Alex Suykov
-Alexander Monakov
-Andre McCurdy
-Andrew Kelley
-Anthony G. Basile
-Aric Belsito
-Arvid Picciani
-Bartosz Brachaczek
-Benjamin Peterson
-Bobby Bingham
-Boris Brezillon
-Brent Cook
-Chris Spiegel
-Clément Vasseur
-Daniel Micay
-Daniel Sabogal
-Daurnimator
-David Carlier
-David Edelsohn
-Denys Vlasenko
-Dmitry Ivanov
-Dmitry V. Levin
-Drew DeVault
-Emil Renner Berthing
-Fangrui Song
-Felix Fietkau
-Felix Janda
-Gianluca Anzolin
-Hauke Mehrtens
-He X
-Hiltjo Posthuma
-Isaac Dunham
-Jaydeep Patil
-Jens Gustedt
-Jeremy Huntwork
-Jo-Philipp Wich
-Joakim Sindholt
-John Spencer
-Julien Ramseier
-Justin Cormack
-Kaarle Ritvanen
-Khem Raj
-Kylie McClain
-Leah Neukirchen
-Luca Barbato
-Luka Perkov
-Lynn Ochs
-M Farkas-Dyck (Strake)
-Mahesh Bodapati
-Markus Wichmann
-Masanori Ogino
-Michael Clark
-Michael Forney
-Mikhail Kremnyov
-Natanael Copa
-Nicholas J. Kain
-orc
-Pascal Cuoq
-Patrick Oppenlander
-Petr Hosek
-Petr Skocik
-Pierre Carrier
-Reini Urban
-Rich Felker
-Richard Pennington
-Ryan Fairfax
-Samuel Holland
-Segev Finer
-Shiz
-sin
-Solar Designer
-Stefan Kristiansson
-Stefan O'Rear
-Szabolcs Nagy
-Timo Teräs
-Trutz Behn
-Will Dietz
-William Haddon
-William Pitcock
-
-Portions of this software are derived from third-party works licensed
-under terms compatible with the above MIT license:
-
-The TRE regular expression implementation (src/regex/reg* and
-src/regex/tre*) is Copyright © 2001-2008 Ville Laurikari and licensed
-under a 2-clause BSD license (license text in the source files). The
-included version has been heavily modified by Rich Felker in 2012, in
-the interests of size, simplicity, and namespace cleanliness.
-
-Much of the math library code (src/math/* and src/complex/*) is
-Copyright © 1993,2004 Sun Microsystems or
-Copyright © 2003-2011 David Schultz or
-Copyright © 2003-2009 Steven G. Kargl or
-Copyright © 2003-2009 Bruce D. Evans or
-Copyright © 2008 Stephen L. Moshier or
-Copyright © 2017-2018 Arm Limited
-and labelled as such in comments in the individual source files. All
-have been licensed under extremely permissive terms.
-
-The ARM memcpy code (src/string/arm/memcpy.S) is Copyright © 2008
-The Android Open Source Project and is licensed under a two-clause BSD
-license. It was taken from Bionic libc, used on Android.
-
-The AArch64 memcpy and memset code (src/string/aarch64/*) are
-Copyright © 1999-2019, Arm Limited.
-
-The implementation of DES for crypt (src/crypt/crypt_des.c) is
-Copyright © 1994 David Burren. It is licensed under a BSD license.
-
-The implementation of blowfish crypt (src/crypt/crypt_blowfish.c) was
-originally written by Solar Designer and placed into the public
-domain. The code also comes with a fallback permissive license for use
-in jurisdictions that may not recognize the public domain.
-
-The smoothsort implementation (src/stdlib/qsort.c) is Copyright © 2011
-Lynn Ochs and is licensed under an MIT-style license.
-
-The x86_64 port was written by Nicholas J. Kain and is licensed under
-the standard MIT terms.
-
-The mips and microblaze ports were originally written by Richard
-Pennington for use in the ellcc project. The original code was adapted
-by Rich Felker for build system and code conventions during upstream
-integration. It is licensed under the standard MIT terms.
-
-The mips64 port was contributed by Imagination Technologies and is
-licensed under the standard MIT terms.
-
-The powerpc port was also originally written by Richard Pennington,
-and later supplemented and integrated by John Spencer. It is licensed
-under the standard MIT terms.
-
-All other files which have no copyright comments are original works
-produced specifically for use as part of this library, written either
-by Rich Felker, the main author of the library, or by one or more
-contibutors listed above. Details on authorship of individual files
-can be found in the git version control history of the project. The
-omission of copyright and license comments in each file is in the
-interest of source tree size.
-
-In addition, permission is hereby granted for all public header files
-(include/* and arch/*/bits/*) and crt files intended to be linked into
-applications (crt/*, ldso/dlstart.c, and arch/*/crt_arch.h) to omit
-the copyright notice and permission notice otherwise required by the
-license, and to use these files without any requirement of
-attribution. These files include substantial contributions from:
-
-Bobby Bingham
-John Spencer
-Nicholas J. Kain
-Rich Felker
-Richard Pennington
-Stefan Kristiansson
-Szabolcs Nagy
-
-all of whom have explicitly granted such permission.
-
-This file previously contained text expressing a belief that most of
-the files covered by the above exception were sufficiently trivial not
-to be subject to copyright, resulting in confusion over whether it
-negated the permissions granted in the license. In the spirit of
-permissive licensing, and of not having licensing issues being an
-obstacle to adoption, that text has been removed.

locator/tablets.hh

@@ -200,10 +200,7 @@ enum class tablet_repair_incremental_mode : uint8_t {
     disabled,
 };
 
-// FIXME: Incremental repair is disabled by default due to
-// https://github.com/scylladb/scylladb/issues/26041 and
-// https://github.com/scylladb/scylladb/issues/27414
-constexpr tablet_repair_incremental_mode default_tablet_repair_incremental_mode{tablet_repair_incremental_mode::disabled};
+constexpr tablet_repair_incremental_mode default_tablet_repair_incremental_mode{tablet_repair_incremental_mode::incremental};
 
 sstring tablet_repair_incremental_mode_to_string(tablet_repair_incremental_mode);
 tablet_repair_incremental_mode tablet_repair_incremental_mode_from_string(const sstring&);

main.cc

@@ -39,6 +39,7 @@
 #include "api/api_init.hh"
 #include "db/config.hh"
 #include "db/extensions.hh"
+#include "db/legacy_schema_migrator.hh"
 #include "service/storage_service.hh"
 #include "service/migration_manager.hh"
 #include "service/tablet_allocator.hh"
@@ -748,6 +749,8 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
     // inherit Seastar's CPU affinity masks. We want this thread to be free
     // to migrate between CPUs; we think that's what makes the most sense.
     auto rpc_dict_training_worker = utils::alien_worker(startlog, 19, "rpc-dict");
+    // niceness=10 is ~10% of normal process time
+    auto hashing_worker = utils::alien_worker(startlog, 10, "pwd-hash");
 
     return app.run(ac, av, [&] () -> future<int> {
 
@@ -777,7 +780,8 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
         return seastar::async([&app, cfg, ext, &disk_space_monitor_shard0, &cm, &sstm, &db, &qp, &bm, &proxy, &mapreduce_service, &mm, &mm_notifier, &ctx, &opts, &dirs,
                 &prometheus_server, &cf_cache_hitrate_calculator, &load_meter, &feature_service, &gossiper, &snitch,
                 &token_metadata, &erm_factory, &snapshot_ctl, &messaging, &sst_dir_semaphore, &raft_gr, &service_memory_limiter,
-                &repair, &sst_loader, &auth_cache, &ss, &lifecycle_notifier, &stream_manager, &task_manager, &rpc_dict_training_worker, &vector_store_client] {
+                &repair, &sst_loader, &auth_cache, &ss, &lifecycle_notifier, &stream_manager, &task_manager, &rpc_dict_training_worker,
+                &hashing_worker, &vector_store_client] {
           try {
               if (opts.contains("relabel-config-file") && !opts["relabel-config-file"].as<sstring>().empty()) {
                   // calling update_relabel_config_from_file can cause an exception that would stop startup
@@ -1637,7 +1641,7 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
             fd.start(
                 std::ref(fd_pinger), std::ref(fd_clock),
                 service::direct_fd_clock::base::duration{std::chrono::milliseconds{100}}.count(),
-                service::direct_fd_clock::base::duration{std::chrono::milliseconds{cfg->direct_failure_detector_ping_timeout_in_ms()}}.count(), dbcfg.gossip_scheduling_group).get();
+                service::direct_fd_clock::base::duration{std::chrono::milliseconds{cfg->direct_failure_detector_ping_timeout_in_ms()}}.count()).get();
 
             auto stop_fd = defer_verbose_shutdown("direct_failure_detector", [] {
                 fd.stop().get();
@@ -1847,6 +1851,8 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
             group0_client.init().get();
 
             checkpoint(stop_signal, "initializing system schema");
+            // schema migration, if needed, is also done on shard 0
+            db::legacy_schema_migrator::migrate(proxy, db, sys_ks, qp.local()).get();
             db::schema_tables::save_system_schema(qp.local()).get();
             db::schema_tables::recalculate_schema_version(sys_ks, proxy, feature_service.local()).get();
 
@@ -2057,7 +2063,7 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
                 maintenance_auth_config.authenticator_java_name = sstring{auth::allow_all_authenticator_name};
                 maintenance_auth_config.role_manager_java_name = sstring{auth::maintenance_socket_role_manager_name};
 
-                maintenance_auth_service.start(perm_cache_config, std::ref(qp), std::ref(group0_client),  std::ref(mm_notifier), std::ref(mm), maintenance_auth_config, maintenance_socket_enabled::yes, std::ref(auth_cache)).get();
+                maintenance_auth_service.start(perm_cache_config, std::ref(qp), std::ref(group0_client),  std::ref(mm_notifier), std::ref(mm), maintenance_auth_config, maintenance_socket_enabled::yes, std::ref(auth_cache), std::ref(hashing_worker)).get();
 
                 cql_maintenance_server_ctl.emplace(maintenance_auth_service, mm_notifier, gossiper, qp, service_memory_limiter, sl_controller, lifecycle_notifier, *cfg, maintenance_cql_sg_stats_key, maintenance_socket_enabled::yes, dbcfg.statement_scheduling_group);
 
@@ -2333,7 +2339,7 @@ To start the scylla server proper, simply invoke as: scylla server (or just scyl
             auth_config.authenticator_java_name = qualified_authenticator_name;
             auth_config.role_manager_java_name = qualified_role_manager_name;
 
-            auth_service.start(std::move(perm_cache_config), std::ref(qp), std::ref(group0_client), std::ref(mm_notifier), std::ref(mm), auth_config, maintenance_socket_enabled::no, std::ref(auth_cache)).get();
+            auth_service.start(std::move(perm_cache_config), std::ref(qp), std::ref(group0_client), std::ref(mm_notifier), std::ref(mm), auth_config, maintenance_socket_enabled::no, std::ref(auth_cache), std::ref(hashing_worker)).get();
 
             std::any stop_auth_service;
             // Has to be called after node joined the cluster (join_cluster())

message/messaging_service.cc

@@ -686,7 +686,6 @@ static constexpr unsigned do_get_rpc_client_idx(messaging_verb verb) {
     case messaging_verb::RAFT_MODIFY_CONFIG:
     case messaging_verb::RAFT_PULL_SNAPSHOT:
     case messaging_verb::NOTIFY_BANNED:
-    case messaging_verb::DIRECT_FD_PING:
         // See comment above `TOPOLOGY_INDEPENDENT_IDX`.
         // DO NOT put any 'hot' (e.g. data path) verbs in this group,
         // only verbs which are 'rare' and 'cheap'.
@@ -748,6 +747,7 @@ static constexpr unsigned do_get_rpc_client_idx(messaging_verb verb) {
     case messaging_verb::PAXOS_ACCEPT:
     case messaging_verb::PAXOS_LEARN:
     case messaging_verb::PAXOS_PRUNE:
+    case messaging_verb::DIRECT_FD_PING:
         return 2;
     case messaging_verb::MUTATION_DONE:
     case messaging_verb::MUTATION_FAILED:

mutation/partition_version.cc

@@ -575,15 +575,10 @@ utils::coroutine partition_entry::apply_to_incomplete(const schema& s,
                         }
                         res.row.set_range_tombstone(cur.range_tombstone_for_row() + src_cur.range_tombstone());
 
-                        if (need_preempt()) {
-                            lb = position_in_partition(cur.position());
-                            ++tracker.get_stats().rows_covered_by_range_tombstones_from_memtable;
-                            return stop_iteration::no;
-                        }
-
                         // FIXME: Compact the row
                         ++tracker.get_stats().rows_covered_by_range_tombstones_from_memtable;
                         cur.next();
+                        // FIXME: preempt
                     }
                 }
                 {

pgo/profiles/aarch64/profile.profdata.xz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3cbe2dd05945f8fb76ebce2ea70864063d2b282c4d5080af1f290ead43321ab3
-size 6444732
+oid sha256:80a47fe93866989aaf7e949168fcd308e95841e78c976a61f9eac20bfdd34d96
+size 6448960

raft/tracker.cc

@@ -46,7 +46,7 @@ bool follower_progress::is_stray_reject(const append_reply::rejected& rejected)
         // any reject during snapshot transfer is stray one
         return true;
     default:
-        SCYLLA_ASSERT(false);
+        scylla_assert(false, "invalid follower_progress state: {}", static_cast<int>(state));
     }
     return false;
 }
@@ -87,7 +87,7 @@ bool follower_progress::can_send_to() {
         // before starting to sync the log.
         return false;
     }
-    SCYLLA_ASSERT(false);
+    scylla_assert(false, "invalid follower_progress state in can_send_to: {}", static_cast<int>(state));
     return false;
 }
 

replica/compaction_group.hh

@@ -297,17 +297,17 @@ public:
 
     const dht::token_range& token_range() const noexcept;
 
-    size_t memtable_count() const;
+    size_t memtable_count() const noexcept;
 
     const compaction_group_ptr& main_compaction_group() const noexcept;
     const std::vector<compaction_group_ptr>& split_ready_compaction_groups() const;
     compaction_group_ptr& select_compaction_group(locator::tablet_range_side) noexcept;
 
-    uint64_t live_disk_space_used() const;
+    uint64_t live_disk_space_used() const noexcept;
 
-    void for_each_compaction_group(std::function<void(const compaction_group_ptr&)> action) const;
-    utils::small_vector<compaction_group_ptr, 3> compaction_groups();
-    utils::small_vector<const_compaction_group_ptr, 3> compaction_groups() const;
+    void for_each_compaction_group(std::function<void(const compaction_group_ptr&)> action) const noexcept;
+    utils::small_vector<compaction_group_ptr, 3> compaction_groups() noexcept;
+    utils::small_vector<const_compaction_group_ptr, 3> compaction_groups() const noexcept;
 
     utils::small_vector<compaction_group_ptr, 3> split_unready_groups() const;
     bool split_unready_groups_are_empty() const;
@@ -430,7 +430,7 @@ public:
     virtual storage_group& storage_group_for_token(dht::token) const = 0;
     virtual utils::chunked_vector<storage_group_ptr> storage_groups_for_token_range(dht::token_range tr) const = 0;
 
-    virtual locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const = 0;
+    virtual locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const noexcept = 0;
     virtual bool all_storage_groups_split() = 0;
     virtual future<> split_all_storage_groups(tasks::task_info tablet_split_task_info) = 0;
     virtual future<> maybe_split_compaction_group_of(size_t idx) = 0;

replica/database.hh

@@ -1133,7 +1133,7 @@ public:
 
     // The tablet filter is used to not double account migrating tablets, so it's important that
     // only one of pending or leaving replica is accounted based on current migration stage.
-    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const;
+    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const noexcept;
 
     const db::view::stats& get_view_stats() const {
         return _view_stats;

replica/distributed_loader.cc

@@ -234,12 +234,18 @@ distributed_loader::get_sstables_from_upload_dir(sharded<replica::database>& db,
 }
 
 future<std::tuple<table_id, std::vector<std::vector<sstables::shared_sstable>>>>
-distributed_loader::get_sstables_from_object_store(sharded<replica::database>& db, sstring ks, sstring cf, std::vector<sstring> sstables, sstring endpoint, sstring type, sstring bucket, sstring prefix, sstables::sstable_open_config cfg, std::function<seastar::abort_source*()> get_abort_src) {
-    return get_sstables_from(db, ks, cf, cfg, [bucket, endpoint, type, prefix, sstables=std::move(sstables), &get_abort_src] (auto& global_table, auto& directory) {
+distributed_loader::get_sstables_from_object_store(sharded<replica::database>& db, sstring ks, sstring cf, std::vector<sstring> sstables, sstring endpoint, sstring bucket, sstring prefix, sstables::sstable_open_config cfg, std::function<seastar::abort_source*()> get_abort_src) {
+    return get_sstables_from(db, ks, cf, cfg, [bucket, endpoint, prefix, sstables=std::move(sstables), &get_abort_src, &db] (auto& global_table, auto& directory) {
         return directory.start(global_table.as_sharded_parameter(),
-            sharded_parameter([bucket, endpoint, type, prefix, &get_abort_src] {
+            sharded_parameter([bucket, endpoint, prefix, &get_abort_src, &db] {
+                auto eps = db.local().get_config().object_storage_endpoints() 
+                    | std::views::filter([&endpoint](auto& ep) { return ep.key() == endpoint; })
+                    ;
+                if (eps.empty()) {
+                    throw std::invalid_argument(fmt::format("Undefined endpoint {}", endpoint));
+                }
                 seastar::abort_source* as = get_abort_src ? get_abort_src() : nullptr;
-                auto opts = data_dictionary::make_object_storage_options(endpoint, type, bucket, prefix, as);
+                auto opts = data_dictionary::make_object_storage_options(endpoint, eps.front().type(), bucket, prefix, as);
                 return make_lw_shared<const data_dictionary::storage_options>(std::move(opts));
             }),
             sstables,

replica/distributed_loader.hh

@@ -92,7 +92,7 @@ public:
     static future<std::tuple<table_id, std::vector<std::vector<sstables::shared_sstable>>>>
             get_sstables_from_upload_dir(sharded<replica::database>& db, sstring ks, sstring cf, sstables::sstable_open_config cfg);
     static future<std::tuple<table_id, std::vector<std::vector<sstables::shared_sstable>>>>
-            get_sstables_from_object_store(sharded<replica::database>& db, sstring ks, sstring cf, std::vector<sstring> sstables, sstring endpoint, sstring type, sstring bucket, sstring prefix, sstables::sstable_open_config cfg, std::function<seastar::abort_source*()> = {});
+            get_sstables_from_object_store(sharded<replica::database>& db, sstring ks, sstring cf, std::vector<sstring> sstables, sstring endpoint, sstring bucket, sstring prefix, sstables::sstable_open_config cfg, std::function<seastar::abort_source*()> = {});
     static future<> process_upload_dir(sharded<replica::database>& db, sharded<db::view::view_builder>& vb, sharded<db::view::view_building_worker>& vbw, sstring ks_name, sstring cf_name, bool skip_cleanup, bool skip_reshape);
 };
 

replica/table.cc

@@ -708,7 +708,7 @@ public:
         return *_single_sg;
     }
 
-    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)>) const override {
+    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)>) const noexcept override {
         return locator::combined_load_stats{
             .table_ls = locator::table_load_stats{
                             .size_in_bytes = _single_sg->live_disk_space_used(),
@@ -874,7 +874,7 @@ public:
         return storage_group_for_id(storage_group_of(token).first);
     }
 
-    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const override;
+    locator::combined_load_stats table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const noexcept override;
     bool all_storage_groups_split() override;
     future<> split_all_storage_groups(tasks::task_info tablet_split_task_info) override;
     future<> maybe_split_compaction_group_of(size_t idx) override;
@@ -922,7 +922,7 @@ compaction_group_ptr& storage_group::select_compaction_group(locator::tablet_ran
     return _main_cg;
 }
 
-void storage_group::for_each_compaction_group(std::function<void(const compaction_group_ptr&)> action) const {
+void storage_group::for_each_compaction_group(std::function<void(const compaction_group_ptr&)> action) const noexcept {
     action(_main_cg);
     for (auto& cg : _merging_groups) {
         action(cg);
@@ -932,7 +932,7 @@ void storage_group::for_each_compaction_group(std::function<void(const compactio
     }
 }
 
-utils::small_vector<compaction_group_ptr, 3> storage_group::compaction_groups() {
+utils::small_vector<compaction_group_ptr, 3> storage_group::compaction_groups() noexcept {
     utils::small_vector<compaction_group_ptr, 3> cgs;
     for_each_compaction_group([&cgs] (const compaction_group_ptr& cg) {
         cgs.push_back(cg);
@@ -940,7 +940,7 @@ utils::small_vector<compaction_group_ptr, 3> storage_group::compaction_groups()
     return cgs;
 }
 
-utils::small_vector<const_compaction_group_ptr, 3> storage_group::compaction_groups() const {
+utils::small_vector<const_compaction_group_ptr, 3> storage_group::compaction_groups() const noexcept {
     utils::small_vector<const_compaction_group_ptr, 3> cgs;
     for_each_compaction_group([&cgs] (const compaction_group_ptr& cg) {
         cgs.push_back(cg);
@@ -1890,7 +1890,7 @@ sstables::file_size_stats compaction_group::live_disk_space_used_full_stats() co
     return _main_sstables->get_file_size_stats() + _maintenance_sstables->get_file_size_stats();
 }
 
-uint64_t storage_group::live_disk_space_used() const {
+uint64_t storage_group::live_disk_space_used() const noexcept {
     auto cgs = const_cast<storage_group&>(*this).compaction_groups();
     return std::ranges::fold_left(cgs | std::views::transform(std::mem_fn(&compaction_group::live_disk_space_used)), uint64_t(0), std::plus{});
 }
@@ -2813,7 +2813,7 @@ void table::on_flush_timer() {
     });
 }
 
-locator::combined_load_stats tablet_storage_group_manager::table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const {
+locator::combined_load_stats tablet_storage_group_manager::table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const noexcept {
     locator::table_load_stats table_stats;
     table_stats.split_ready_seq_number = _split_ready_seq_number;
 
@@ -2836,7 +2836,7 @@ locator::combined_load_stats tablet_storage_group_manager::table_load_stats(std:
     };
 }
 
-locator::combined_load_stats table::table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const {
+locator::combined_load_stats table::table_load_stats(std::function<bool(const locator::tablet_map&, locator::global_tablet_id)> tablet_filter) const noexcept {
     return _sg_manager->table_load_stats(std::move(tablet_filter));
 }
 
@@ -3453,7 +3453,7 @@ size_t compaction_group::memtable_count() const noexcept {
     return _memtables->size();
 }
 
-size_t storage_group::memtable_count() const {
+size_t storage_group::memtable_count() const noexcept {
     return std::ranges::fold_left(compaction_groups() | std::views::transform(std::mem_fn(&compaction_group::memtable_count)), size_t(0), std::plus{});
 }
 

scripts/get_description.py

@@ -22,7 +22,7 @@ format_match = re.compile(r'\s*(?:seastar::)?format\(\s*"([^"]+)"\s*,\s*(.*)\s*'
 def handle_error(message, strict=True, verbose_mode=False):
     if strict:
         print(f"[ERROR] {message}")
-        exit(1)
+        exit(-1)
     elif verbose_mode:
         print(f"[WARNING] {message}")
 
@@ -180,11 +180,12 @@ def get_metrics_from_file(file_name, prefix, metrics_information, verb=None, str
     groups = {}
     if clean_name in metrics_information:
         if (isinstance(metrics_information[clean_name], str) and metrics_information[clean_name] == "skip") or "skip" in metrics_information[clean_name]:
-            return {}
+            exit(0)
     param_mapping =  metrics_information[clean_name]["params"] if clean_name in metrics_information and "params" in metrics_information[clean_name] else {}
     groups = metrics_information[clean_name]["groups"] if clean_name in metrics_information and "groups" in metrics_information[clean_name] else {}
 
     metrics = {}
+    multi_line = False
     names = undefined
     typ = undefined
     line_number = 0;

scripts/metrics-config.yml

@@ -1,6 +1,6 @@
 "cdc/log.cc":
   params:
-    cdc_group_name: "cdc"
+    cdc_group_name: cdc
     part_name;suffix: [["static_row", "total"],["clustering_row", "total"], ["map", "total"], ["set", "total"], ["list", "total"], ["udt", "total"], ["range_tombstone", "total"],["partition_delete", "total"],["row_delete", "total"], ["static_row", "failed"],["clustering_row", "failed"], ["map", "failed"], ["set", "failed"], ["list", "failed"], ["udt", "failed"], ["range_tombstone", "failed"],["partition_delete", "failed"],["row_delete", "failed"]]
     kind: ["total", "failed"]
 "db/commitlog/commitlog.cc":
@@ -9,7 +9,7 @@
         "cfg.max_active_flushes": "cfg.max_active_flushes"
 "cql3/query_processor.cc":
     groups:
-        "80": "query_processor"
+        "80": query_processor
 "replica/dirty_memory_manager.cc":
   params:
     namestr: ["regular", "system"]
@@ -19,11 +19,10 @@
 "replica/database.cc":
   params:
     "_dirty_memory_manager.throttle_threshold()": "throttle threshold"
-"seastar/apps/metrics_tester/metrics_tester.cc": "skip"
-"seastar/tests/unit/metrics_test.cc": "skip"
-"seastar/tests/unit/metrics_tester.cc": "skip"
-"seastar/tests/unit/prometheus_http_test.cc": "skip"
-"seastar/tests/unit/prometheus_text_test.cc": "skip"
+"seastar/apps/metrics_tester/metrics_tester.cc": skip
+"seastar/tests/unit/metrics_test.cc": skip
+"seastar/tests/unit/metrics_tester.cc": skip
+"seastar/tests/unit/prometheus_http_test.cc": skip
 "service/storage_proxy.cc":
   params:
     COORDINATOR_STATS_CATEGORY: "storage_proxy_coordinator"
@@ -33,25 +32,25 @@
     _short_description_prefix: ["total_write_attempts", "write_errors", "background_replica_writes_failed", "read_repair_write_attempts"]
     _long_description_prefix: ["total number of write requests", "number of write requests that failed", "background_replica_writes_failed", "number of write operations in a read repair context"]
     _category: "storage_proxy_coordinator"
-"thrift/server.cc": "skip"
+"thrift/server.cc": skip
 "tracing/tracing.cc":
   params:
     "max_pending_trace_records + write_event_records_threshold": "max_pending_trace_records + write_event_records_threshold"
 "transport/server.cc":
   groups:
-    "200": "transport"
+    "200": transport
   params:
     "_config.max_request_size": "max_request_size"
-"seastar/src/net/dpdk.cc": "skip"
+"seastar/src/net/dpdk.cc": skip
 "db/hints/manager.cc":
     params:
         "group_name": ["hints_for_views_manager", "hints_manager"]
 "seastar/src/core/execution_stage.cc":
     groups:
-        "100": "execution_stages"
+        "100": execution_stages
 "seastar/src/core/fair_queue.cc":
     groups:
-        "300": "io_queue"
+        "300": io_queue
 "seastar/src/net/net.cc":
     params:
         _stats_plugin_name: ["stats_plugin_name"]

scripts/pull_github_pr.sh

@@ -38,9 +38,8 @@ for required in jq curl; do
 	fi
 done
 
+FORCE=0
 ALLOW_SUBMODULE=0
-ALLOW_UNSTABLE=0
-ALLOW_ANY_BRANCH=0
 
 function print_usage {
 cat << EOF
@@ -61,18 +60,12 @@ Options:
 -h
     Print this help message and exit.
 
---allow-submodule
-    Allow a PR to update a submudule
-
---allow-unstable
+--force
+    Do not check current branch to be next*
     Do not check jenkins job status
 
---allow-any-branch
-    Merge PR even if target branch is not next
-
---force
-    Sets all above --allow-* options
-
+--allow-submodule
+    Allow a PR to update a submudule
 EOF
 }
 
@@ -80,23 +73,13 @@ while [[ $# -gt 0 ]]
 do
     case $1 in
         "--force"|"-f")
-            ALLOW_UNSTABLE=1
-            ALLOW_SUBMODULE=1
-            ALLOW_ANY_BRANCH=1
+            FORCE=1
             shift 1
             ;;
         --allow-submodule)
             ALLOW_SUBMODULE=1
             shift
             ;;
-        --allow-unstable)
-            ALLOW_UNSTABLE=1
-            shift
-            ;;
-        --allow-any-branch)
-            ALLOW_ANY_BRANCH=1
-            shift
-            ;;
         +([0-9]))
             PR_NUM=$1
             shift 1
@@ -164,7 +147,7 @@ check_jenkins_job_status() {
   fi
 }
 
-if [[ $ALLOW_UNSTABLE -eq 0 ]]; then
+if [[ $FORCE -eq 0 ]]; then
   check_jenkins_job_status
 fi
 
@@ -196,19 +179,17 @@ echo -n "Fetching full name of author $PR_LOGIN... "
 USER_NAME=$(curl -s "https://api.github.com/users/$PR_LOGIN" | jq -r .name)
 echo "$USER_NAME"
 
-if [[ $ALLOW_ANY_BRANCH -eq 0 ]]; then
-    BASE_BRANCH=$(jq -r .base.ref <<< $PR_DATA)
-    CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
-    TARGET_BASE="unknown"
-    if [[ ${BASE_BRANCH} == master ]]; then
-        TARGET_BASE="next"
-    elif [[ ${BASE_BRANCH}  == branch-* ]]; then
-        TARGET_BASE=${BASE_BRANCH//branch/next}
-    fi
-    if [[ "${CURRENT_BRANCH}" != "${TARGET_BASE}" ]]; then
-        echo "Merging into wrong next, want ${TARGET_BASE}, have ${CURRENT_BRANCH}. Use --allow-any-branch or --force to skip this check"
-        exit 1
-    fi
+BASE_BRANCH=$(jq -r .base.ref <<< $PR_DATA)
+CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
+TARGET_BASE="unknown"
+if [[ ${BASE_BRANCH} == master ]]; then
+    TARGET_BASE="next"
+elif [[ ${BASE_BRANCH}  == branch-* ]]; then
+    TARGET_BASE=${BASE_BRANCH//branch/next}
+fi
+if [[ "${CURRENT_BRANCH}" != "${TARGET_BASE}" ]]; then
+    echo "Merging into wrong next, want ${TARGET_BASE}, have ${CURRENT_BRANCH}"
+    exit 1
 fi
 
 git fetch "$REMOTE" pull/$PR_NUM/head

service/direct_failure_detector/failure_detector.cc

@@ -6,7 +6,6 @@
  * SPDX-License-Identifier: LicenseRef-ScyllaDB-Source-Available-1.0
  */
 
-#include "seastar/core/scheduling.hh"
 #include "utils/assert.hh"
 #include <unordered_set>
 
@@ -18,7 +17,6 @@
 #include <seastar/core/condition-variable.hh>
 #include <seastar/coroutine/parallel_for_each.hh>
 #include <seastar/util/defer.hh>
-#include <seastar/coroutine/switch_to.hh>
 
 #include "utils/log.hh"
 
@@ -120,7 +118,7 @@ struct failure_detector::impl {
 
     // Fetches endpoint updates from _endpoint_queue and performs the add/remove operation.
     // Runs on shard 0 only.
-    future<> update_endpoint_fiber(seastar::scheduling_group sg);
+    future<> update_endpoint_fiber();
     future<> _update_endpoint_fiber = make_ready_future<>();
 
     // Workers running on this shard.
@@ -142,7 +140,7 @@ struct failure_detector::impl {
     // The unregistering process requires cross-shard operations which we perform on this fiber.
     future<> _destroy_subscriptions = make_ready_future<>();
 
-    impl(failure_detector& parent, pinger&, clock&, clock::interval_t ping_period, clock::interval_t ping_timeout, seastar::scheduling_group sg);
+    impl(failure_detector& parent, pinger&, clock&, clock::interval_t ping_period, clock::interval_t ping_timeout);
     ~impl();
 
     // Inform update_endpoint_fiber() about an added/removed endpoint.
@@ -179,19 +177,19 @@ struct failure_detector::impl {
 };
 
 failure_detector::failure_detector(
-    pinger& pinger, clock& clock, clock::interval_t ping_period, clock::interval_t ping_timeout, seastar::scheduling_group sg)
-        : _impl(std::make_unique<impl>(*this, pinger, clock, ping_period, ping_timeout, sg))
+    pinger& pinger, clock& clock, clock::interval_t ping_period, clock::interval_t ping_timeout)
+        : _impl(std::make_unique<impl>(*this, pinger, clock, ping_period, ping_timeout))
 {}
 
 failure_detector::impl::impl(
-    failure_detector& parent, pinger& pinger, clock& clock, clock::interval_t ping_period, clock::interval_t ping_timeout, seastar::scheduling_group sg)
+    failure_detector& parent, pinger& pinger, clock& clock, clock::interval_t ping_period, clock::interval_t ping_timeout)
         : _parent(parent), _pinger(pinger), _clock(clock), _ping_period(ping_period), _ping_timeout(ping_timeout) {
     if (this_shard_id() != 0) {
         return;
     }
 
     _num_workers.resize(smp::count, 0);
-    _update_endpoint_fiber = update_endpoint_fiber(sg);
+    _update_endpoint_fiber = update_endpoint_fiber();
 }
 
 void failure_detector::impl::send_update_endpoint(pinger::endpoint_id ep, endpoint_update update) {
@@ -207,9 +205,9 @@ void failure_detector::impl::send_update_endpoint(pinger::endpoint_id ep, endpoi
     _endpoint_changed.signal();
 }
 
-future<> failure_detector::impl::update_endpoint_fiber(seastar::scheduling_group sg) {
+future<> failure_detector::impl::update_endpoint_fiber() {
     SCYLLA_ASSERT(this_shard_id() == 0);
-    co_await coroutine::switch_to(sg);
+
     while (true) {
         co_await _endpoint_changed.wait([this] { return !_endpoint_updates.empty(); });
 
@@ -482,7 +480,7 @@ static future<bool> ping_with_timeout(pinger::endpoint_id id, clock::timepoint_t
         }
     });
 
-    auto f = pinger.ping(id, timeout, timeout_as, c);
+    auto f = pinger.ping(id, timeout_as);
     auto sleep_and_abort = [] (clock::timepoint_t timeout, abort_source& timeout_as, clock& c) -> future<> {
         co_await c.sleep_until(timeout, timeout_as).then_wrapped([&timeout_as] (auto&& f) {
             // Avoid throwing if sleep was aborted.

service/direct_failure_detector/failure_detector.hh

@@ -19,6 +19,26 @@ class abort_source;
 
 namespace direct_failure_detector {
 
+class pinger {
+public:
+    // Opaque endpoint ID.
+    // A specific implementation of `pinger` maps those IDs to 'real' addresses.
+    using endpoint_id = utils::UUID;
+
+    // Send a message to `ep` and wait until it responds.
+    // The wait can be aborted using `as`.
+    // Abort should be signalized with `abort_requested_exception`.
+    //
+    // If the ping fails in an expected way (e.g. the endpoint is down and refuses to connect),
+    // returns `false`. If it succeeds, returns `true`.
+    virtual future<bool> ping(endpoint_id ep, abort_source& as) = 0;
+
+protected:
+    // The `pinger` object must not be destroyed through the `pinger` interface.
+    // `failure_detector` does not take ownership of `pinger`, only a non-owning reference.
+    ~pinger() = default;
+};
+
 // A clock that uses abstract units to measure time.
 // The implementation is responsible for periodically advancing the clock.
 //
@@ -40,33 +60,12 @@ public:
     // Aborts should be signalized using `seastar::sleep_aborted`.
     virtual future<> sleep_until(timepoint_t tp, abort_source& as) = 0;
 
-    virtual std::chrono::milliseconds to_milliseconds(timepoint_t tp) const = 0;
 protected:
     // The `clock` object must not be destroyed through the `clock` interface.
     // `failure_detector` does not take ownership of `clock`, only a non-owning reference.
     ~clock() = default;
 };
 
-class pinger {
-public:
-    // Opaque endpoint ID.
-    // A specific implementation of `pinger` maps those IDs to 'real' addresses.
-    using endpoint_id = utils::UUID;
-
-    // Send a message to `ep` and wait until it responds.
-    // The wait can be aborted using `as`.
-    // Abort should be signalized with `abort_requested_exception`.
-    //
-    // If the ping fails in an expected way (e.g. the endpoint is down and refuses to connect),
-    // returns `false`. If it succeeds, returns `true`.
-    virtual future<bool> ping(endpoint_id ep, clock::timepoint_t timeout, abort_source& as, clock& c) = 0;
-
-protected:
-    // The `pinger` object must not be destroyed through the `pinger` interface.
-    // `failure_detector` does not take ownership of `pinger`, only a non-owning reference.
-    ~pinger() = default;
-};
-
 class listener {
 public:
     // Called when an endpoint in the detected set (added by `failure_detector::add_endpoint`) responds to a ping
@@ -128,10 +127,7 @@ public:
 
         // Duration after which a ping is aborted, so that next ping can be started
         // (pings are sent sequentially).
-        clock::interval_t ping_timeout,
-
-        // Scheduling group used for fibers inside the failure detector.
-        seastar::scheduling_group sg
+        clock::interval_t ping_timeout
     );
 
     ~failure_detector();

service/raft/raft_group_registry.cc

@@ -18,7 +18,6 @@
 #include "utils/error_injection.hh"
 #include "seastar/core/shared_future.hh"
 
-#include <chrono>
 #include <seastar/core/coroutine.hh>
 #include <seastar/core/when_all.hh>
 #include <seastar/core/sleep.hh>
@@ -203,11 +202,8 @@ void raft_group_registry::init_rpc_verbs() {
     });
 
     ser::raft_rpc_verbs::register_direct_fd_ping(&_ms,
-            [this] (const rpc::client_info&, rpc::opt_time_point timeout, raft::server_id dst) -> future<direct_fd_ping_reply> {
-
-        if (timeout && *timeout <= netw::messaging_service::clock_type::now()) {
-            throw timed_out_error{};
-        }
+            [this] (const rpc::client_info&, raft::server_id dst) -> future<direct_fd_ping_reply> {
+        // XXX: update address map here as well?
 
         if (_my_id != dst) {
             return make_ready_future<direct_fd_ping_reply>(direct_fd_ping_reply {
@@ -217,10 +213,19 @@ void raft_group_registry::init_rpc_verbs() {
             });
         }
 
-        return make_ready_future<direct_fd_ping_reply>(direct_fd_ping_reply {
-            .result = service::group_liveness_info{
-                .group0_alive = _group0_is_alive,
+        return container().invoke_on(0, [] (raft_group_registry& me) -> future<direct_fd_ping_reply> {
+            bool group0_alive = false;
+            if (me._group0_id) {
+                auto* group0_server = me.find_server(*me._group0_id);
+                if (group0_server && group0_server->is_alive()) {
+                    group0_alive = true;
+                }
             }
+            co_return direct_fd_ping_reply {
+                .result = service::group_liveness_info{
+                    .group0_alive = group0_alive,
+                }
+            };
         });
     });
 }
@@ -375,12 +380,6 @@ future<> raft_group_registry::start_server_for_group(raft_server_for_group new_g
         co_await server.abort();
         std::rethrow_exception(ex);
     }
-
-    if (gid == _group0_id) {
-        co_await container().invoke_on_all([] (raft_group_registry& rg) {
-            rg._group0_is_alive = true;
-        });
-    }
 }
 
 future<> raft_group_registry::abort_server(raft::group_id gid, sstring reason) {
@@ -390,18 +389,14 @@ future<> raft_group_registry::abort_server(raft::group_id gid, sstring reason) {
     if (const auto it = _servers.find(gid); it != _servers.end()) {
         auto& [gid, s] = *it;
         if (!s.aborted) {
-            if (gid == _group0_id) {
-                co_await container().invoke_on_all([] (raft_group_registry& rg) {
-                    rg._group0_is_alive = false;
-                });
-            }
             s.aborted = s.server->abort(std::move(reason))
                 .handle_exception([gid] (std::exception_ptr ex) {
                     rslog.warn("Failed to abort raft group server {}: {}", gid, ex);
                 });
         }
-        co_await s.aborted->get_future();
+        return s.aborted->get_future();
     }
+    return make_ready_future<>();
 }
 
 unsigned raft_group_registry::shard_for_group(const raft::group_id& gid) const {
@@ -522,13 +517,11 @@ future<> raft_server_with_timeouts::read_barrier(seastar::abort_source* as, std:
     }, "read_barrier", as, timeout);
 }
 
-future<bool> direct_fd_pinger::ping(direct_failure_detector::pinger::endpoint_id id, direct_failure_detector::clock::timepoint_t timeout, abort_source& as, direct_failure_detector::clock& c) {
+future<bool> direct_fd_pinger::ping(direct_failure_detector::pinger::endpoint_id id, abort_source& as) {
     auto dst_id = raft::server_id{id};
 
     try {
-        std::chrono::milliseconds timeout_ms = c.to_milliseconds(timeout);
-        netw::messaging_service::clock_type::time_point deadline = netw::messaging_service::clock_type::now() + timeout_ms;
-        auto reply = co_await ser::raft_rpc_verbs::send_direct_fd_ping(&_ms, locator::host_id{id}, deadline, as, dst_id);
+        auto reply = co_await ser::raft_rpc_verbs::send_direct_fd_ping(&_ms, locator::host_id{id}, as, dst_id);
         if (auto* wrong_dst = std::get_if<wrong_destination>(&reply.result)) {
             // FIXME: after moving to host_id based verbs we will not get `wrong_destination`
             //        any more since the connection will fail
@@ -561,11 +554,4 @@ future<> direct_fd_clock::sleep_until(direct_failure_detector::clock::timepoint_
     return sleep_abortable(t - n, as);
 }
 
-std::chrono::milliseconds direct_fd_clock::to_milliseconds(direct_failure_detector::clock::timepoint_t tp) const {
-    auto t = base::time_point{base::duration{tp}};
-    auto n = base::now();
-    return std::chrono::duration_cast<std::chrono::milliseconds>(t - n);
-}
-
-
 } // end of namespace service

service/raft/raft_group_registry.hh

@@ -127,7 +127,6 @@ private:
     // My Raft ID. Shared between different Raft groups.
     raft::server_id _my_id;
 
-    bool _group0_is_alive = false;
 public:
     raft_group_registry(raft::server_id my_id, netw::messaging_service& ms,
             direct_failure_detector::failure_detector& fd);
@@ -182,9 +181,6 @@ public:
     unsigned shard_for_group(const raft::group_id& gid) const;
     shared_ptr<raft::failure_detector> failure_detector();
     direct_failure_detector::failure_detector& direct_fd() { return _direct_fd; }
-    bool is_group0_alive() const {
-        return _group0_is_alive;
-    }
 };
 
 // Implementation of `direct_failure_detector::pinger` which uses DIRECT_FD_PING verb for pinging.
@@ -202,7 +198,7 @@ public:
     direct_fd_pinger(const direct_fd_pinger&) = delete;
     direct_fd_pinger(direct_fd_pinger&&) = delete;
 
-    future<bool> ping(direct_failure_detector::pinger::endpoint_id id, direct_failure_detector::clock::timepoint_t timeout, abort_source& as, direct_failure_detector::clock& c) override;
+    future<bool> ping(direct_failure_detector::pinger::endpoint_id id, abort_source& as) override;
 };
 
 // XXX: find a better place to put this?
@@ -211,7 +207,6 @@ struct direct_fd_clock : public direct_failure_detector::clock {
 
     direct_failure_detector::clock::timepoint_t now() noexcept override;
     future<> sleep_until(direct_failure_detector::clock::timepoint_t tp, abort_source& as) override;
-    std::chrono::milliseconds to_milliseconds(direct_failure_detector::clock::timepoint_t tp) const override;
 };
 
 } // end of namespace service

service/storage_proxy.cc

@@ -6688,11 +6688,10 @@ storage_proxy::do_query_with_paxos(schema_ptr s,
         }
     };
 
-    auto request = std::make_unique<read_cas_request>();
-    auto* request_ptr = request.get();
+    auto request = seastar::make_shared<read_cas_request>();
 
-    return cas(std::move(s), std::move(cas_shard), *request_ptr, cmd, std::move(partition_ranges), std::move(query_options),
-            cl, db::consistency_level::ANY, timeout, cas_timeout, false).then([request = std::move(request)] (bool is_applied) mutable {
+    return cas(std::move(s), std::move(cas_shard), request, cmd, std::move(partition_ranges), std::move(query_options),
+            cl, db::consistency_level::ANY, timeout, cas_timeout, false).then([request] (bool is_applied) mutable {
         return make_ready_future<coordinator_query_result>(std::move(request->res));
     });
 }
@@ -6755,13 +6754,11 @@ static mutation_write_failure_exception read_failure_to_write(read_failure_excep
  * NOTE: `cmd` argument can be nullptr, in which case it's guaranteed that this function would not perform
  * any reads of committed values (in case user of the function is not interested in them).
  *
- * NOTE: The `request` object must be guaranteed to be alive until the returned future is resolved.
- *
  * WARNING: the function must be called on a shard that owns the key cas() operates on.
  * The cas_shard must be created *before* selecting the shard, to protect against
  * concurrent tablet migrations.
  */
-future<bool> storage_proxy::cas(schema_ptr schema, cas_shard cas_shard, cas_request& request, lw_shared_ptr<query::read_command> cmd,
+future<bool> storage_proxy::cas(schema_ptr schema, cas_shard cas_shard, shared_ptr<cas_request> request, lw_shared_ptr<query::read_command> cmd,
         dht::partition_range_vector partition_ranges, storage_proxy::coordinator_query_options query_options,
         db::consistency_level cl_for_paxos, db::consistency_level cl_for_learn,
         clock_type::time_point write_timeout, clock_type::time_point cas_timeout, bool write, cdc::per_request_options cdc_opts) {
@@ -6862,7 +6859,7 @@ future<bool> storage_proxy::cas(schema_ptr schema, cas_shard cas_shard, cas_requ
                 qr = std::move(cqr.query_result);
             }
 
-            auto mutation = request.apply(std::move(qr), cmd->slice, utils::UUID_gen::micros_timestamp(ballot), cdc_opts);
+            auto mutation = request->apply(std::move(qr), cmd->slice, utils::UUID_gen::micros_timestamp(ballot), cdc_opts);
             condition_met = true;
             if (!mutation) {
                 if (write) {

service/storage_proxy.hh

@@ -829,7 +829,7 @@ public:
         clock_type::time_point timeout,
         tracing::trace_state_ptr trace_state = nullptr);
 
-    future<bool> cas(schema_ptr schema, cas_shard cas_shard, cas_request& request, lw_shared_ptr<query::read_command> cmd,
+    future<bool> cas(schema_ptr schema, cas_shard cas_shard, shared_ptr<cas_request> request, lw_shared_ptr<query::read_command> cmd,
             dht::partition_range_vector partition_ranges, coordinator_query_options query_options,
             db::consistency_level cl_for_paxos, db::consistency_level cl_for_learn,
             clock_type::time_point write_timeout, clock_type::time_point cas_timeout, bool write = true, cdc::per_request_options cdc_opts = {});

service/storage_service.cc

@@ -600,7 +600,7 @@ future<storage_service::nodes_to_notify_after_sync> storage_service::sync_raft_t
             co_await update_topology_change_info(tmptr, ::format("{} {}/{}", rs.state, id, ip));
             break;
         case node_state::replacing: {
-            SCYLLA_ASSERT(_topology_state_machine._topology.req_param.contains(id));
+            scylla_assert(_topology_state_machine._topology.req_param.contains(id));
             auto replaced_id = std::get<replace_param>(_topology_state_machine._topology.req_param[id]).replaced_id;
             auto existing_ip = _address_map.find(locator::host_id{replaced_id.uuid()});
             const auto replaced_host_id = locator::host_id(replaced_id.uuid());
@@ -688,7 +688,7 @@ future<> storage_service::notify_nodes_after_sync(nodes_to_notify_after_sync&& n
 future<> storage_service::topology_state_load(state_change_hint hint) {
 #ifdef SEASTAR_DEBUG
     static bool running = false;
-    SCYLLA_ASSERT(!running); // The function is not re-entrant
+    scylla_assert(!running); // The function is not re-entrant
     auto d = defer([] {
         running = false;
     });
@@ -854,7 +854,7 @@ future<> storage_service::topology_state_load(state_change_hint hint) {
 }
 
 future<> storage_service::topology_transition(state_change_hint hint) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     co_await topology_state_load(std::move(hint)); // reload new state
 
     _topology_state_machine.event.broadcast();
@@ -898,7 +898,7 @@ future<> storage_service::view_building_state_load() {
 }
 
 future<> storage_service::view_building_transition() {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     co_await view_building_state_load();
 
     _view_building_state_machine.event.broadcast();
@@ -966,7 +966,7 @@ future<> storage_service::merge_topology_snapshot(raft_snapshot snp) {
 }
 
 future<> storage_service::update_service_levels_cache(qos::update_both_cache_levels update_only_effective_cache, qos::query_context ctx) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     if (_sl_controller.local().is_v2()) {
         // Skip cache update unless the topology upgrade is done
         co_await _sl_controller.local().update_cache(update_only_effective_cache, ctx);
@@ -1520,7 +1520,7 @@ future<> storage_service::update_topology_with_local_metadata(raft::server& raft
 }
 
 future<> storage_service::start_upgrade_to_raft_topology() {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
 
     if (_topology_state_machine._topology.upgrade_state != topology::upgrade_state_type::not_upgraded) {
         co_return;
@@ -1572,7 +1572,7 @@ future<> storage_service::start_upgrade_to_raft_topology() {
 }
 
 topology::upgrade_state_type storage_service::get_topology_upgrade_state() const {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     return _topology_state_machine._topology.upgrade_state;
 }
 
@@ -1841,7 +1841,7 @@ future<> storage_service::join_topology(sharded<service::storage_proxy>& proxy,
         slogger.info("Nodes {} are alive", get_sync_nodes());
     }
 
-    SCYLLA_ASSERT(_group0);
+    scylla_assert(_group0);
 
     join_node_request_params join_params {
         .host_id = _group0->load_my_id(),
@@ -2083,7 +2083,7 @@ future<> storage_service::join_topology(sharded<service::storage_proxy>& proxy,
 
     if (!_sys_ks.local().bootstrap_complete()) {
         // If we're not bootstrapping then we shouldn't have chosen a CDC streams timestamp yet.
-        SCYLLA_ASSERT(should_bootstrap() || !cdc_gen_id);
+        scylla_assert(should_bootstrap() || !cdc_gen_id);
 
         // Don't try rewriting CDC stream description tables.
         // See cdc.md design notes, `Streams description table V1 and rewriting` section, for explanation.
@@ -2167,7 +2167,7 @@ future<> storage_service::join_topology(sharded<service::storage_proxy>& proxy,
         throw std::runtime_error(err);
     }
 
-    SCYLLA_ASSERT(_group0);
+    scylla_assert(_group0);
     co_await _group0->finish_setup_after_join(*this, _qp, _migration_manager.local(), false);
     co_await _cdc_gens.local().after_join(std::move(cdc_gen_id));
 
@@ -2192,7 +2192,7 @@ future<> storage_service::join_topology(sharded<service::storage_proxy>& proxy,
 }
 
 future<> storage_service::track_upgrade_progress_to_topology_coordinator(sharded<service::storage_proxy>& proxy) {
-    SCYLLA_ASSERT(_group0);
+    scylla_assert(_group0);
 
     while (true) {
         _group0_as.check();
@@ -2316,7 +2316,7 @@ future<> storage_service::bootstrap(std::unordered_set<token>& bootstrap_tokens,
 
             // After we pick a generation timestamp, we start gossiping it, and we stick with it.
             // We don't do any other generation switches (unless we crash before complecting bootstrap).
-            SCYLLA_ASSERT(!cdc_gen_id);
+            scylla_assert(!cdc_gen_id);
 
             cdc_gen_id = _cdc_gens.local().legacy_make_new_generation(bootstrap_tokens, !is_first_node()).get();
 
@@ -2349,9 +2349,9 @@ future<> storage_service::bootstrap(std::unordered_set<token>& bootstrap_tokens,
             slogger.debug("Removing replaced endpoint {} from system.peers", replace_addr);
             _sys_ks.local().remove_endpoint(replace_addr).get();
 
-            SCYLLA_ASSERT(replaced_host_id);
+            scylla_assert(replaced_host_id);
             auto raft_id = raft::server_id{replaced_host_id.uuid()};
-            SCYLLA_ASSERT(_group0);
+            scylla_assert(_group0);
             bool raft_available = _group0->wait_for_raft().get();
             if (raft_available) {
                 slogger.info("Replace: removing {}/{} from group 0...", replace_addr, raft_id);
@@ -3000,7 +3000,7 @@ future<> storage_service::stop_transport() {
 }
 
 future<> storage_service::drain_on_shutdown() {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     return (_operation_mode == mode::DRAINING || _operation_mode == mode::DRAINED) ?
         _drain_finished.get_future() : do_drain();
 }
@@ -3025,7 +3025,7 @@ bool storage_service::is_topology_coordinator_enabled() const {
 
 future<> storage_service::join_cluster(sharded<service::storage_proxy>& proxy,
         start_hint_manager start_hm, gms::generation_type new_generation) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
 
     if (_sys_ks.local().was_decommissioned()) {
         auto msg = sstring("This node was decommissioned and will not rejoin the ring unless "
@@ -3225,7 +3225,7 @@ future<> storage_service::join_cluster(sharded<service::storage_proxy>& proxy,
 }
 
 future<token_metadata_change> storage_service::prepare_token_metadata_change(mutable_token_metadata_ptr tmptr, const schema_getter& schema_getter) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     std::exception_ptr ex;
     token_metadata_change change;
 
@@ -3992,7 +3992,7 @@ future<> storage_service::decommission() {
                 slogger.info("DECOMMISSIONING: starts");
                 ctl.req.leaving_nodes = std::list<gms::inet_address>{endpoint};
 
-                SCYLLA_ASSERT(ss._group0);
+                scylla_assert(ss._group0);
                 bool raft_available = ss._group0->wait_for_raft().get();
 
                 try {
@@ -4044,7 +4044,7 @@ future<> storage_service::decommission() {
 
                     if (raft_available && left_token_ring) {
                         slogger.info("decommission[{}]: leaving Raft group 0", uuid);
-                        SCYLLA_ASSERT(ss._group0);
+                        scylla_assert(ss._group0);
                         ss._group0->leave_group0().get();
                         slogger.info("decommission[{}]: left Raft group 0", uuid);
                     }
@@ -4350,7 +4350,7 @@ future<> storage_service::removenode(locator::host_id host_id, locator::host_id_
             auto stop_ctl = deferred_stop(ctl);
             auto uuid = ctl.uuid();
             const auto& tmptr = ctl.tmptr;
-            SCYLLA_ASSERT(ss._group0);
+            scylla_assert(ss._group0);
             auto raft_id = raft::server_id{host_id.uuid()};
             bool raft_available = ss._group0->wait_for_raft().get();
             bool is_group0_member = raft_available && ss._group0->is_member(raft_id, false);
@@ -4470,7 +4470,7 @@ future<> storage_service::removenode(locator::host_id host_id, locator::host_id_
 }
 
 future<> storage_service::check_and_repair_cdc_streams() {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
 
     if (!_cdc_gens.local_is_initialized()) {
         return make_exception_future<>(std::runtime_error("CDC generation service not initialized yet"));
@@ -5784,7 +5784,7 @@ future<> storage_service::mutate_token_metadata(std::function<future<> (mutable_
 }
 
 future<> storage_service::update_topology_change_info(mutable_token_metadata_ptr tmptr, sstring reason) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
 
     try {
         locator::dc_rack_fn get_dc_rack_by_host_id([this, &tm = *tmptr] (locator::host_id host_id) -> std::optional<locator::endpoint_dc_rack> {
@@ -5831,7 +5831,7 @@ future<> storage_service::keyspace_changed(const sstring& ks_name) {
 }
 
 future<locator::mutable_token_metadata_ptr> storage_service::prepare_tablet_metadata(const locator::tablet_metadata_change_hint& hint, mutable_token_metadata_ptr pending_token_metadata) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     if (hint) {
         co_await replica::update_tablet_metadata(_db.local(), _qp, pending_token_metadata->tablets(), hint);
     } else {
@@ -5955,7 +5955,7 @@ void storage_service::start_tablet_split_monitor() {
 }
 
 future<> storage_service::snitch_reconfigured() {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
     auto& snitch = _snitch.local();
     co_await mutate_token_metadata([&snitch] (mutable_token_metadata_ptr tmptr) -> future<> {
         // re-read local rack and DC info
@@ -6487,8 +6487,8 @@ future<> storage_service::stream_tablet(locator::global_tablet_id tablet) {
     co_await utils::get_local_injector().inject("block_tablet_streaming", [this, &tablet] (auto& handler) -> future<> {
         const auto keyspace = handler.get("keyspace");
         const auto table = handler.get("table");
-        SCYLLA_ASSERT(keyspace);
-        SCYLLA_ASSERT(table);
+        scylla_assert(keyspace);
+        scylla_assert(table);
         auto s = _db.local().find_column_family(tablet.table).schema();
         bool should_block = s->ks_name() == *keyspace && s->cf_name() == *table;
         while (should_block && !handler.poll_for_message() && !_async_gate.is_closed()) {
@@ -7509,7 +7509,7 @@ future<join_node_request_result> storage_service::join_node_request_handler(join
 }
 
 future<join_node_response_result> storage_service::join_node_response_handler(join_node_response_params params) {
-    SCYLLA_ASSERT(this_shard_id() == 0);
+    scylla_assert(this_shard_id() == 0);
 
     // Usually this handler will only run once, but there are some cases where we might get more than one RPC,
     // possibly happening at the same time, e.g.:

service/topology_coordinator.cc

@@ -360,7 +360,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
         auto& topo = _topo_sm._topology;
 
         auto it = topo.find(id);
-        SCYLLA_ASSERT(it);
+        scylla_assert(it);
 
         std::optional<topology_request> req;
         auto rit = topo.requests.find(id);
@@ -2310,7 +2310,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
 
                 switch (node.rs->state) {
                     case node_state::bootstrapping: {
-                        SCYLLA_ASSERT(!node.rs->ring);
+                        scylla_assert(!node.rs->ring);
                         auto num_tokens = std::get<join_param>(node.req_param.value()).num_tokens;
                         auto tokens_string = std::get<join_param>(node.req_param.value()).tokens_string;
 
@@ -2359,11 +2359,11 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
                     }
                         break;
                     case node_state::replacing: {
-                        SCYLLA_ASSERT(!node.rs->ring);
+                        scylla_assert(!node.rs->ring);
                         auto replaced_id = std::get<replace_param>(node.req_param.value()).replaced_id;
                         auto it = _topo_sm._topology.normal_nodes.find(replaced_id);
-                        SCYLLA_ASSERT(it != _topo_sm._topology.normal_nodes.end());
-                        SCYLLA_ASSERT(it->second.ring && it->second.state == node_state::normal);
+                        scylla_assert(it != _topo_sm._topology.normal_nodes.end());
+                        scylla_assert(it->second.ring && it->second.state == node_state::normal);
 
                         topology_mutation_builder builder(node.guard.write_timestamp());
 
@@ -3022,7 +3022,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
                 rtbuilder.set("start_time", db_clock::now());
                 switch (node.request.value()) {
                     case topology_request::join: {
-                        SCYLLA_ASSERT(!node.rs->ring);
+                        scylla_assert(!node.rs->ring);
                         // Write chosen tokens through raft.
                         builder.set_transition_state(topology::transition_state::join_group0)
                                .with_node(node.id)
@@ -3033,7 +3033,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
                         break;
                         }
                     case topology_request::leave:
-                        SCYLLA_ASSERT(node.rs->ring);
+                        scylla_assert(node.rs->ring);
                         // start decommission and put tokens of decommissioning nodes into write_both_read_old state
                         // meaning that reads will go to the replica being decommissioned
                         // but writes will go to new owner as well
@@ -3046,7 +3046,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
                                                        "start decommission");
                         break;
                     case topology_request::remove: {
-                        SCYLLA_ASSERT(node.rs->ring);
+                        scylla_assert(node.rs->ring);
 
                         builder.set_transition_state(topology::transition_state::tablet_draining)
                                .set_version(_topo_sm._topology.version + 1)
@@ -3058,7 +3058,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
                         break;
                         }
                     case topology_request::replace: {
-                        SCYLLA_ASSERT(!node.rs->ring);
+                        scylla_assert(!node.rs->ring);
 
                         builder.set_transition_state(topology::transition_state::join_group0)
                                .with_node(node.id)
@@ -3163,7 +3163,7 @@ class topology_coordinator : public endpoint_lifecycle_subscriber {
 
         auto id = node.id;
 
-        SCYLLA_ASSERT(!_topo_sm._topology.transition_nodes.empty());
+        scylla_assert(!_topo_sm._topology.transition_nodes.empty());
 
         release_node(std::move(node));
 
@@ -4013,7 +4013,7 @@ future<> topology_coordinator::stop() {
         // but let's check all of them because we never reset these holders
         // once they are added as barriers
         for (auto& [stage, barrier]: tablet_state.barriers) {
-            SCYLLA_ASSERT(barrier.has_value());
+            scylla_assert(barrier.has_value());
             co_await stop_background_action(barrier, gid, [stage] { return format("at stage {}", tablet_transition_stage_to_string(stage)); });
         }
 

sstables/compress.cc

@@ -251,7 +251,7 @@ void compression::discard_hidden_options() {
 }
 
 compressor& compression::get_compressor() const {
-    SCYLLA_ASSERT(_compressor);
+    scylla_assert(_compressor);
     return *_compressor.get();
 }
 

sstables/compress.hh

@@ -170,7 +170,7 @@ struct compression {
             const_iterator(const const_iterator& other) = default;
 
             const_iterator& operator=(const const_iterator& other) {
-                SCYLLA_ASSERT(&_offsets == &other._offsets);
+                scylla_assert(&_offsets == &other._offsets);
                 _index = other._index;
                 return *this;
             }

sstables/compressor.cc

@@ -24,6 +24,7 @@
 #include "sstables/sstable_compressor_factory.hh"
 #include "compressor.hh"
 #include "exceptions/exceptions.hh"
+#include "utils/assert.hh"
 #include "utils/config_file_impl.hh"
 #include "utils/class_registrator.hh"
 #include "gms/feature_service.hh"
@@ -295,7 +296,7 @@ size_t zstd_processor::uncompress(const char* input, size_t input_len, char* out
         if (_ddict) {
             return ZSTD_decompress_usingDDict(dctx, output, output_len, input, input_len, _ddict->dict());
         } else {
-            SCYLLA_ASSERT(!_cdict && "Write-only compressor used for reading");
+            scylla_assert(!_cdict && "Write-only compressor used for reading");
             return ZSTD_decompressDCtx(dctx, output, output_len, input, input_len);
         }
     });
@@ -310,7 +311,7 @@ size_t zstd_processor::compress(const char* input, size_t input_len, char* outpu
         if (_cdict) {
             return ZSTD_compress_usingCDict(cctx, output, output_len, input, input_len, _cdict->dict());
         } else {
-            SCYLLA_ASSERT(!_ddict && "Read-only compressor used for writing");
+            scylla_assert(!_ddict && "Read-only compressor used for writing");
             return ZSTD_compressCCtx(cctx, output, output_len, input, input_len, _compression_level);
         }
     });
@@ -627,7 +628,7 @@ size_t lz4_processor::uncompress(const char* input, size_t input_len,
     if (_ddict) {
         ret = LZ4_decompress_safe_usingDict(input, output, input_len, output_len, reinterpret_cast<const char*>(_ddict->raw().data()), _ddict->raw().size());
     } else {
-        SCYLLA_ASSERT(!_cdict && "Write-only compressor used for reading");
+        scylla_assert(!_cdict && "Write-only compressor used for reading");
         ret = LZ4_decompress_safe(input, output, input_len, output_len);
     }
     if (ret < 0) {
@@ -657,7 +658,7 @@ size_t lz4_processor::compress(const char* input, size_t input_len,
             LZ4_resetStream_fast(ctx);
         }
     } else {
-        SCYLLA_ASSERT(!_ddict && "Read-only compressor used for writing");
+        scylla_assert(!_ddict && "Read-only compressor used for writing");
         ret = LZ4_compress_default(input, output + 4, input_len, LZ4_compressBound(input_len));
     }
     if (ret == 0) {
@@ -1268,7 +1269,7 @@ lz4_cdict::~lz4_cdict() {
 }
 
 std::unique_ptr<sstable_compressor_factory> make_sstable_compressor_factory_for_tests_in_thread() {
-    SCYLLA_ASSERT(thread::running_in_thread());
+    scylla_assert(thread::running_in_thread());
     struct wrapper : sstable_compressor_factory {
         using impl = default_sstable_compressor_factory;
         sharded<impl> _impl;

sstables/downsampling.hh

@@ -44,14 +44,14 @@ public:
      * @return A list of `sampling_level` unique indices between 0 and `sampling_level`
      */
     static const std::vector<int>& get_sampling_pattern(int sampling_level) {
-        SCYLLA_ASSERT(sampling_level > 0 && sampling_level <= BASE_SAMPLING_LEVEL);
+        scylla_assert(sampling_level > 0 && sampling_level <= BASE_SAMPLING_LEVEL);
         auto& entry = _sample_pattern_cache[sampling_level-1];
         if (!entry.empty()) {
             return entry;
         }
 
         if (sampling_level <= 1) {
-            SCYLLA_ASSERT(_sample_pattern_cache[0].empty());
+            scylla_assert(_sample_pattern_cache[0].empty());
             _sample_pattern_cache[0].push_back(0);
             return _sample_pattern_cache[0];
         }
@@ -96,7 +96,7 @@ public:
      * @return a list of original indexes for current summary entries
      */
     static const std::vector<int>& get_original_indexes(int sampling_level) {
-        SCYLLA_ASSERT(sampling_level > 0 && sampling_level <= BASE_SAMPLING_LEVEL);
+        scylla_assert(sampling_level > 0 && sampling_level <= BASE_SAMPLING_LEVEL);
         auto& entry = _original_index_cache[sampling_level-1];
         if (!entry.empty()) {
             return entry;
@@ -128,7 +128,7 @@ public:
      * @return the number of partitions before the next index summary entry, inclusive on one end
      */
     static int get_effective_index_interval_after_index(int index, int sampling_level, int min_index_interval) {
-        SCYLLA_ASSERT(index >= -1);
+        scylla_assert(index >= -1);
         const std::vector<int>& original_indexes = get_original_indexes(sampling_level);
         if (index == -1) {
             return original_indexes[0] * min_index_interval;

sstables/exceptions.hh

@@ -31,7 +31,7 @@ public:
 [[noreturn]] void on_parse_error(sstring message, std::optional<component_name> filename);
 [[noreturn, gnu::noinline]] void on_bti_parse_error(uint64_t pos);
 
-// Use this instead of SCYLLA_ASSERT() or assert() in code that is used while parsing SSTables.
+// Use this instead of scylla_assert() or assert() in code that is used while parsing SSTables.
 // SSTables can be corrupted either by ScyllaDB itself or by a freak accident like cosmic background
 // radiation hitting the disk the wrong way. Either way a corrupt SSTable should not bring down the
 // whole server. This method will call on_internal_error() if the condition is false.

sstables/generation_type.hh

@@ -129,7 +129,7 @@ public:
     /// way to determine that is overlapping its partition-ranges with the shard's
     /// owned ranges.
     static bool maybe_owned_by_this_shard(const sstables::generation_type& gen) {
-        SCYLLA_ASSERT(bool(gen));
+        scylla_assert(bool(gen));
         int64_t hint = 0;
         if (gen.is_uuid_based()) {
             hint = std::hash<utils::UUID>{}(gen.as_uuid());

sstables/index_reader.hh

@@ -57,10 +57,7 @@ public:
     index_list indexes;
 
     index_consumer(logalloc::region& r, schema_ptr s)
-        : _s(s)
-        , _alloc_section(abstract_formatter([s] (fmt::format_context& ctx) {
-            fmt::format_to(ctx.out(), "index_consumer {}.{}", s->ks_name(), s->cf_name());
-        }))
+        : _s(std::move(s))
         , _region(r)
     { }
 
@@ -788,9 +785,6 @@ public:
                                                       _sstable->manager().get_cache_tracker().region(),
                                                       _sstable->manager().get_cache_tracker().get_partition_index_cache_stats()))
         , _index_cache(caching ? *_sstable->_index_cache : *_local_index_cache)
-        , _alloc_section(abstract_formatter([sst = _sstable] (fmt::format_context& ctx) {
-            fmt::format_to(ctx.out(), "index_reader {}", sst->get_filename());
-        }))
         , _region(_sstable->manager().get_cache_tracker().region())
         , _use_caching(caching)
         , _single_page_read(single_partition_read) // all entries for a given partition are within a single page

sstables/mx/bsearch_clustered_cursor.hh

@@ -284,9 +284,6 @@ public:
         , _clustering_parser(s, permit, _ctr.clustering_column_value_fix_legths(), true)
         , _block_parser(s, permit, _ctr.clustering_column_value_fix_legths())
         , _permit(std::move(permit))
-        , _as(abstract_formatter([s] (fmt::format_context& ctx) {
-            fmt::format_to(ctx.out(), "cached_promoted_index {}.{}", s.ks_name(), s.cf_name());
-        }))
     { }
 
     ~cached_promoted_index() {

sstables/mx/writer.cc

@@ -91,7 +91,7 @@ public:
         {}
 
         void increment() {
-            SCYLLA_ASSERT(_range);
+            scylla_assert(_range);
             if (!_range->next()) {
                 _range = nullptr;
             }
@@ -102,7 +102,7 @@ public:
         }
 
         const ValueType dereference() const {
-            SCYLLA_ASSERT(_range);
+            scylla_assert(_range);
             return _range->get_value();
         }
 
@@ -153,7 +153,7 @@ public:
         auto limit = std::min(_serialization_limit_size, _offset + clustering_block::max_block_size);
 
         _current_block = {};
-        SCYLLA_ASSERT (_offset % clustering_block::max_block_size == 0);
+        scylla_assert (_offset % clustering_block::max_block_size == 0);
         while (_offset < limit) {
             auto shift = _offset % clustering_block::max_block_size;
             if (_offset < _prefix.size(_schema)) {
@@ -280,7 +280,7 @@ public:
                     ++_current_index;
                 }
             } else {
-                SCYLLA_ASSERT(_mode == encoding_mode::large_encode_missing);
+                scylla_assert(_mode == encoding_mode::large_encode_missing);
                 while (_current_index < total_size) {
                     auto cell = _row.find_cell(_columns[_current_index].get().id);
                     if (!cell) {
@@ -1180,7 +1180,7 @@ void writer::write_cell(bytes_ostream& writer, const clustering_key_prefix* clus
 
     if (cdef.is_counter()) {
         if (!is_deleted) {
-            SCYLLA_ASSERT(!cell.is_counter_update());
+            scylla_assert(!cell.is_counter_update());
             auto ccv = counter_cell_view(cell);
             write_counter_value(ccv, writer, _sst.get_version(), [] (bytes_ostream& out, uint32_t value) {
                 return write_vint(out, value);
@@ -1489,7 +1489,7 @@ template <typename W>
 requires Writer<W>
 static void write_clustering_prefix(sstable_version_types v, W& writer, bound_kind_m kind,
     const schema& s, const clustering_key_prefix& clustering) {
-    SCYLLA_ASSERT(kind != bound_kind_m::static_clustering);
+    scylla_assert(kind != bound_kind_m::static_clustering);
     write(v, writer, kind);
     auto is_ephemerally_full = ephemerally_full_prefix{s.is_compact_table()};
     if (kind != bound_kind_m::clustering) {

sstables/partition_index_cache.hh

@@ -59,7 +59,7 @@ private:
                 // Live entry_ptr should keep the entry alive, except when the entry failed on loading.
                 // In that case, entry_ptr holders are not supposed to use the pointer, so it's safe
                 // to nullify those entry_ptrs.
-                SCYLLA_ASSERT(!ready());
+                scylla_assert(!ready());
             }
         }
 

sstables/sstable_directory.cc

@@ -496,7 +496,7 @@ sstable_directory::move_foreign_sstables(sharded<sstable_directory>& source_dire
             return make_ready_future<>();
         }
         // Should be empty, since an SSTable that belongs to this shard is not remote.
-        SCYLLA_ASSERT(shard_id != this_shard_id());
+        scylla_assert(shard_id != this_shard_id());
         dirlog.debug("Moving {} unshared SSTables of {}.{} to shard {} ", info_vec.size(), _schema->ks_name(), _schema->cf_name(), shard_id);
         return source_directory.invoke_on(shard_id, &sstables::sstable_directory::load_foreign_sstables, std::move(info_vec));
     });
@@ -540,7 +540,7 @@ sstable_directory::collect_output_unshared_sstables(std::vector<sstables::shared
     dirlog.debug("Collecting {} output SSTables (remote={})", resharded_sstables.size(), remote_ok);
     return parallel_for_each(std::move(resharded_sstables), [this, remote_ok] (sstables::shared_sstable sst) {
         auto shards = sst->get_shards_for_this_sstable();
-        SCYLLA_ASSERT(shards.size() == 1);
+        scylla_assert(shards.size() == 1);
         auto shard = shards[0];
 
         if (shard == this_shard_id()) {

sstables/sstable_set.cc

@@ -283,7 +283,7 @@ bool partitioned_sstable_set::store_as_unleveled(const shared_sstable& sst) cons
         }
         sstlog.info("SSTable {}, as_unleveled={}, expect_unleveled={}, sst_tr={}, overlap_ratio={}",
             sst->generation(), as_unleveled, expect_unleveled, sst_tr, dht::overlap_ratio(_token_range, sst_tr));
-        SCYLLA_ASSERT(as_unleveled == expect_unleveled);
+        scylla_assert(as_unleveled == expect_unleveled);
     });
 
     return as_unleveled;
@@ -712,8 +712,8 @@ public:
 
         // by !empty(bound) and `_it` invariant:
         //      _it != _end, _it->first <= bound, and filter(*_it->second) == true
-        SCYLLA_ASSERT(_cmp(_it->first, bound) <= 0);
-        // we don't SCYLLA_ASSERT(filter(*_it->second)) due to the requirement that `filter` is called at most once for each sstable
+        scylla_assert(_cmp(_it->first, bound) <= 0);
+        // we don't scylla_assert(filter(*_it->second)) due to the requirement that `filter` is called at most once for each sstable
 
         // Find all sstables with the same position as `_it` (they form a contiguous range in the container).
         auto next = std::find_if(std::next(_it), _end, [this] (const value_t& v) { return _cmp(v.first, _it->first) != 0; });
@@ -1301,7 +1301,7 @@ sstable_set::create_single_key_sstable_reader(
         mutation_reader::forwarding fwd_mr,
         const sstable_predicate& predicate,
         sstables::integrity_check integrity) const {
-    SCYLLA_ASSERT(pr.is_singular() && pr.start()->value().has_key());
+    scylla_assert(pr.is_singular() && pr.start()->value().has_key());
     return _impl->create_single_key_sstable_reader(cf, std::move(schema),
             std::move(permit), sstable_histogram, pr, slice, std::move(trace_state), fwd, fwd_mr, predicate, integrity);
 }
@@ -1408,7 +1408,7 @@ sstable_set::make_local_shard_sstable_reader(
 {
     auto reader_factory_fn = [s, permit, &slice, trace_state, fwd, fwd_mr, &monitor_generator, &predicate, integrity]
             (shared_sstable& sst, const dht::partition_range& pr) mutable {
-        SCYLLA_ASSERT(!sst->is_shared());
+        scylla_assert(!sst->is_shared());
         if (!predicate(*sst)) {
             return make_empty_mutation_reader(s, permit);
         }

sstables/sstables.cc

@@ -36,6 +36,7 @@
 
 #include "utils/error_injection.hh"
 #include "utils/to_string.hh"
+#include "utils/assert.hh"
 #include "data_dictionary/storage_options.hh"
 #include "dht/sharder.hh"
 #include "writer.hh"
@@ -2485,6 +2486,11 @@ void sstable::validate_originating_host_id() const {
         }
         return;
     }
+
+    if (*originating_host_id != local_host_id) {
+        // FIXME refrain from throwing an exception because of #10148
+        sstlog.warn("Host id {} does not match local host id {} while validating SSTable: {}. Load foreign SSTables via the upload dir instead.", *originating_host_id, local_host_id, get_filename());
+    }
 }
 
 sstring sstable::component_basename(const sstring& ks, const sstring& cf, version_types version, generation_type generation,
@@ -4156,7 +4162,7 @@ future<data_sink> file_io_extension::wrap_sink(const sstable& sst, component_typ
 }
 
 future<data_source> file_io_extension::wrap_source(const sstable& sst, component_type c, data_source) {
-    SCYLLA_ASSERT(0 && "You are not supposed to get here, file_io_extension::wrap_source() is not implemented");
+    scylla_assert(0 && "You are not supposed to get here, file_io_extension::wrap_source() is not implemented");
 }
 
 namespace trie {

sstables/sstables_manager.cc

@@ -55,9 +55,9 @@ sstables_manager::sstables_manager(
 }
 
 sstables_manager::~sstables_manager() {
-    SCYLLA_ASSERT(_closing);
-    SCYLLA_ASSERT(_active.empty());
-    SCYLLA_ASSERT(_undergoing_close.empty());
+    scylla_assert(_closing);
+    scylla_assert(_active.empty());
+    scylla_assert(_undergoing_close.empty());
 }
 
 void sstables_manager::subscribe(sstables_manager_event_handler& handler) {
@@ -135,17 +135,13 @@ future<> storage_manager::update_config(const db::config& cfg) {
     co_return;
 }
 
-auto storage_manager::get_endpoint(const sstring& endpoint) -> object_storage_endpoint& {
+shared_ptr<sstables::object_storage_client> storage_manager::get_endpoint_client(sstring endpoint) {
     auto found = _object_storage_endpoints.find(endpoint);
     if (found == _object_storage_endpoints.end()) {
         smlogger.error("unable to find {} in configured object-storage endpoints", endpoint);
         throw std::invalid_argument(format("endpoint {} not found", endpoint));
     }
-    return found->second;
-}
-
-shared_ptr<sstables::object_storage_client> storage_manager::get_endpoint_client(sstring endpoint) {
-    auto& ep = get_endpoint(endpoint);
+    auto& ep = found->second;
     if (ep.client == nullptr) {
         ep.client = make_object_storage_client(ep.cfg, _object_storage_clients_memory, [&ct = container()] (std::string ep) {
             return ct.local().get_endpoint_client(ep);
@@ -154,10 +150,6 @@ shared_ptr<sstables::object_storage_client> storage_manager::get_endpoint_client
     return ep.client;
 }
 
-sstring storage_manager::get_endpoint_type(sstring endpoint) {
-    return get_endpoint(endpoint).cfg.type();
-}
-
 bool storage_manager::is_known_endpoint(sstring endpoint) const {
     return _object_storage_endpoints.contains(endpoint);
 }