Otherwise, it may be only a temporary situation due to lack of
candidates, and may be unnecessarily alerting.
Also, print node stats to allow assessing how bad the situation is on
the spot. Those stats can hint to a cause of imbalance, if balancing
is per-DC and racks have different capacity.
(cherry picked from commit 4a161bff2d)
Saves on lookup cost, esp. for candidate evaluation. This showed up in
perf profile in the past.
Also, lays the ground for splitting stats per rack.
(cherry picked from commit 0d090aa47b)
Load-balancing can be now per-rack instead of per-DC. So just printing
"in DC" is confusing. If we're balancing a rack, we should print which
rack is that.
(cherry picked from commit f2b0146f0f)
Before:
load_balancer - Prepared 1 migration plans, out of which there were 1 tablet migration(s) and 0 resize decision(s) and 0 tablet repair(s) and 0 rack-list colocation(s)
After:
load_balancer - Prepared plan: migrations: 1
We print only stats about elements which are present.
(cherry picked from commit df32318f66)
Previously we only inspected std::system_error inside
std::nested_exception to support a specific TLS-related failure
mode. However, nested exceptions may contain any type, including
other restartable (retryable) errors. This change unwraps one
nested exception per iteration and re-applies all known handlers
until a match is found or the chain is exhausted.
Closesscylladb/scylladb#28240
(cherry picked from commit cb2aa85cf5)
Closesscylladb/scylladb#28345
Commit 0156e97560 ("storage_proxy: cas: reject for
tablets-enabled tables") marked a bunch of LWT tests as
XFAIL with tablets enabled, pending resolution of #18066.
But since that event is now in the past, we undo the XFAIL
markings (or in some cases, use an any-keyspace fixture
instead of a vnodes-only fixture).
Ref #18066.
Closesscylladb/scylladb#28336
(cherry picked from commit ec70cea2a1)
Closesscylladb/scylladb#28365
The class in question has internal implementations of in-memory data_sink_impl and data_source_impl. In seastar there's generic implementation of the same facilities. From the "code re-use" perspective it makes sense to use both. TODO-s in Scylla code supports that.
Using newer seastar facilities, not backporting.
Closesscylladb/scylladb#28321
* github.com:scylladb/scylladb:
sstable: Replace buffer_data_sink_impl with seastar::util::basic_memory_data_sink
sstables: Use seastar::util::as_input_stream() and remove buffer_data_source_impl
There are two checks for live endpoints performed in test_gossiper.py,
but one of those sits in test_gossiper_unreachable_endpoints somehow.
This patch moves live endpoints check into live endpoints test.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
Closesscylladb/scylladb#28224
This series implements rescoring algorithm.
Index options allowing to enable this functionality were introduced in earlier PR https://github.com/scylladb/scylladb/pull/28165.
When Vector Index has enabled quantization, Vector Store uses reduced vector representation to save memory, but it may degrade correctness of ANN queries. For quantized index we can enable rescoring algorithm, which recalculates similarity score from full vector representation stored in Scylla and reorder returned result set.
It works also with oversampling - we fetch more candidates from Vector Store, rescore them at Scylla and return only requested number of results.
Example:
Creating a Vector Index with Rescoring
```sql
-- Create a table with a vector column
CREATE TABLE ks.products (
id int PRIMARY KEY,
embedding vector<float, 128>
);
-- Create a vector index with rescoring enabled
CREATE INDEX products_embedding_idx ON ks.products (embedding)
USING 'vector_index'
WITH OPTIONS = {
'similarity_function': 'cosine',
'quantization': 'i8',
'oversampling': '2.0',
'rescoring': 'true'
};
```
1. **Quantization** (`i8`) compresses vectors in the index, reducing memory usage but introducing precision loss in distance calculations
2. **Oversampling** (`2.0`) retrieves 2× more candidates than requested from the vector store (e.g., `LIMIT 10` fetches 20 candidates)
3. **Rescoring** (`true`) recalculates similarity scores using full-precision (`f32`) vectors from the base table and re-ranks results
Query example:
```sql
-- Find 10 most similar products
SELECT id, similarity_cosine(embedding, [0.1, 0.2, ...]) AS score
FROM ks.products
ORDER BY embedding ANN OF [0.1, 0.2, ...]
LIMIT 10;
```
With rescoring enabled, the query:
1. Fetches 20 candidates from the quantized index (due to oversampling=2.0)
2. Reads full-precision embeddings from the base table
3. Recalculates similarity scores with full precision
4. Re-ranks and returns the top 10 results
In this implementation we use CQL similarity function implementation to calculate new score values and use them in post query ordering. We add that column manually to selection, but it has to be removed from the final response.
Follow-up https://github.com/scylladb/scylladb/pull/28165
Fixes https://scylladb.atlassian.net/browse/SCYLLADB-83
New feature - doesn't need backport.
Closesscylladb/scylladb#27769
* github.com:scylladb/scylladb:
vector_index: rescoring: Fetch oversampled rows
vector_index: rescoring: Sort by similarity column
select_statement: Modify `needs_post_query_ordering` condition
vector_index: rescoring: Add hidden similarity score column
vector_index: Refactor extracting ANN query information
The storage_proxy::stop() is not called by main (it is commented out due to #293), so the corresponding message injection is never hit. When the test releases paxos_state_learn_after_mutate, shutdown may already be in progress or even completed by the time we try to trigger the storage_proxy::stop injection, which makes the test flaky.
Fix this by completely removing the storage_proxy::stop injection. The injection is not required for test correctness. Shutdown must wait for the background LWT learn to finish, which is released via the paxos_state_learn_after_mutate injection. The shutdown process blocks on in-flight HTTP requests through seastar::httpd::http_server::stop and its _task_gate, so the HTTP request that releases paxos_state_learn_after_mutate is guaranteed to complete before the node is shut down.
Fixesscylladb/scylladb#28260
backport: 2025.4, the `test_lwt_shutdown` test was introduced in this version
Closesscylladb/scylladb#28315
* https://github.com/scylladb/scylladb:
storage_proxy: drop stop() method
test_lwt_shutdown: fix flakiness by removing storage_proxy::stop injection
It was observed twice that the test times out in debug mode.
Fix by increasing the timeout.
The test never expects a timeout, so increasing it won't increase
the test duration.
Fixes#28028Closesscylladb/scylladb#28272
- Pass pytest request fixture into coro_task (used for scylla_tmp_dir
and core dump path)
- Rename duplicate `test_sstable_summary` that runs sstable-index-cache
to `test_sstable_index_cache` so both tests are collected
Refs https://github.com/scylladb/scylladb/issues/22501Closesscylladb/scylladb#28286
The former accumulates sstable writer writes into a vector of temporary
buffers. In seastar there's a generic memory data sink that provides a
sink to accumulate stream of bytes into any container.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
The latter is used to wrap vector of buffers into an input_stream.
Seastar already provides the very same functionality with the
convenience as_input_stream() helper.
Signed-off-by: Pavel Emelyanov <xemul@scylladb.com>
In 8df61f6d99 we changed the requirements for creating materialized
views and MV-based indexes - instead of requiring the
rf_rack_valid_keyspaces flag to be set, we now require the keyspace to
be RF-rack-valid at the time of creation, and it is enforced to remain
RF-rack-valid while the MV exists. This validation is done in the cql
create view/index statements.
The same should be done also for alternator - when creating a table with
GSI or LSI, or when adding a GSI to an existing table, previously we
required the flag rf_rack_valid_keyspaces to be set. Now we change it to
instead check if the keyspace is RF-rack-valid, and if not the operation
fails with an appropriate error.
Fixes https://github.com/scylladb/scylladb/issues/28214
backport to 2025.4 to add RF-rack-valid enforcements in alternator
Closesscylladb/scylladb#28154
* github.com:scylladb/scylladb:
locator: document the exception type of assert_rf_rack_valid_keyspace
alternator: don't require rf_rack flag for indexes, validate instead
When read or write operations are performed on a DC with RF=0 with LOCAL_QUORUM
or LOCAL_ONE consistency level, Cassandra throws `Unavailable` exception.
Scylla allowed such read operations and failed write operations with a cryptic:
"broken promise" error. This occured because the initial availability
check passed (quorum of 0 requires 0 replicas), but execution failed
later when no replicas existed to process the mutation.
This patch adds an explicit RF=0 validation for LOCAL_ONE and LOCAL_QUORUM that
throws before attempting operation execution.
The change also requires `test_query_dc_with_rf_0_does_not_crash_db` to be
upgraded. This testcase was asserting somewhat similar scenario, but wasn't
taking into account the whole matrix of combinations:
- scenarios: successful vs unsuccesful operation outcome
- local consistency levels: LOCAL_QUORUM & LOCAL_ONE
- operations: SELECT (read) & INSERT (write)
and so it's been extended to cover both the pre-existing and the current issues
and the whole matrix of combinations.
Fixes: scylladb/scylladb#27893
A minor change, no need to backport.
Closesscylladb/scylladb#27894
* github.com:scylladb/scylladb:
db: fail reads and writes with local consistencty level to a DC with RF=0
db: consistency_level: split `local_quorum_for()`
db: consistency_level: fix nrs -> nts abbreviation
storage_proxy::stop() is not called by main (it is commented out due to #293),
so the corresponding message injection is never hit. When the test releases
paxos_state_learn_after_mutate, shutdown may already be in progress or even
completed by the time we try to trigger the storage_proxy::stop injection,
which makes the test flaky.
Fix this by completely removing the storage_proxy::stop injection.
The injection is not required for test correctness. Shutdown must wait for the
background LWT learn to finish, which is released via the
paxos_state_learn_after_mutate injection.
The shutdown process blocks on in-flight api HTTP requests through
seastar::httpd::http_server::stop and its _task_gate, so the
shutdown will not prevent the HTTP request that released the
paxos_state_learn_after_mutate from completing successfully.
Fixesscylladb/scylladb#28260
During rewrite --extra-scylla-cmdline-options was missed and it was not passed to the tests that are using pytest. The issue that there were no possibility to pass these parameters via cmd to the Scylla, while tests were not affected because they were using the parameters from the yaml file.
This PR fixes this issue so it will be easier to modify the Scylla start parameters without modifying code.
No backport needed, only framework enhancement.
Closesscylladb/scylladb#28156
* github.com:scylladb/scylladb:
test.py: do not crash when there is no boost log
test.py: pass correctly extra cmd line arguments
Before this change, the test function `_verify_tasks_processed_metrics`
verified that after service level reconfiguration, a given number of
`scylla_scheduler_tasks_processed` were processed by a given scheduling
group. Moreover, the check verified that another scheduling group
didn't process a high number of requests. The second check was vulnerable
to flakiness, because sometimes additional load caused extensive work
in the second scheduling group (e.g. password hashing in `sl:driver`
due to new connections being created).
To avoid test failures, this commit changes which metric is verified:
instead of `scylla_scheduler_tasks_processed`, the metric
`scylla_transport_cql_requests_count` is checked. This prevents similar
problems, because there is no reason for a high number of
requests to be processed by the second scheduling group. Moreover,
it allows decreasing the number of requests that are sent for
verification, and thus speeds up the test.
Fixes: scylladb/scylladb#27715Closesscylladb/scylladb#28318
In this PR we add a basic implementation of the strongly-consistent tables:
* generate raft group id when a strongly-consistent table is created
* persist it into system.tables table
* start raft groups on replicas when a strongly-consistent tablet_map reaches them
* add strongly-consistent version of the storage_proxy, with the `query` and `mutate` methods
* the `mutate` method submits a command to the tablets raft group, the query method reads the data with `raft.read_barrier()`
* strongly-consistent versions of the `select_statement` and `modification_statement` are added
* a basic `test_strong_consistency.py/test_basic_write_read` is added which to check that we can write and read data in a strongly consistent fashion.
Limitations:
* for now the strongly consistent tables can have tablets only on shard zero. This is because we (ab/re) use the existing raft system tables which live only on shard0. In the next PRs we'll create separate tables for the new tablets raft groups.
* No Scylla-side proxying - the test has to figure out who is the leader and submit the command to the right node. This will be fixed separately.
* No tablet balancing -- migration/split/merges require separate complicated code.
The new behavior is hidden behind `STRONGLY_CONSISTENT_TABLES` feature, which is enabled when the `STRONGLY_CONSISTENT_TABLES` experimental feature flag is set.
Requirements, specs and general overview of the feature can be found [here](https://scylladb.atlassian.net/wiki/spaces/RND/pages/91422722/Strong+Consistency). Short term implementation plan is [here](https://docs.google.com/document/d/1afKeeHaCkKxER7IThHkaAQlh2JWpbqhFLIQ3CzmiXhI/edit?tab=t.0#heading=h.thkorgfek290)
One can check the strongly consistent writes and reads locally via cqlsh:
scylla.yaml:
```
experimental_features:
- strongly-consistent-tables
```
cqlsh:
```
CREATE KEYSPACE IF NOT EXISTS my_ks WITH replication = {'class': 'NetworkTopologyStrategy', 'replication_factor': 1} AND tablets = {'initial': 1} AND consistency = 'local';
CREATE TABLE my_ks.test (pk int PRIMARY KEY, c int);
INSERT INTO my_ks.test (pk, c) VALUES (10, 20);
SELECT * FROM my_ks.test WHERE pk = 10;
```
Fixes SCYLLADB-34
Fixes SCYLLADB-32
Fixes SCYLLADB-31
Fixes SCYLLADB-33
Fixes SCYLLADB-56
backport: no need
Closesscylladb/scylladb#27614
* https://github.com/scylladb/scylladb:
test_encryption: capture stderr
test/cluster: add test_strong_consistency.py
raft_group_registry: disable metrics for non-0 groups
strong consistency: implement select_statement::do_execute()
cql: add select_statement.cc
strong consistency: implement coordinator::query()
cql: add modification_statement
cql: add statement_helpers
strong consistency: implement coordinator::mutate()
raft.hh: make server::wait_for_leader() public
strong_consistency: add coordinator
modification_statement: make get_timeout public
strong_consistency: add groups_manager
strong_consistency: add state_machine and raft_command
table: add get_max_timestamp_for_tablet
tablets: generate raft group_id-s for new table
tablet_replication_strategy: add consistency field
tablets: add raft_group_id
modification_statement: remove virtual where it's not needed
modification_statement: inline prepare_statement()
system_keyspace: disable tablet_balancing for strongly_consistent_tables
cql: rename strongly_consistent statements to broadcast statements
This patch adds links to the Vector Search documentation that is hosted
together with Scylla Cloud docs to the CQL documentation.
It also make the note about supported capabilities consistent and
removes the experimental label as the feature is GAed.
Fixes: SCYLLADB-371
Closesscylladb/scylladb#28312
The function assert_rf_rack_valid_keyspace uses the exception type
std::invalid_argument when the RF-rack validation fails. Document it and
change all callers to catch this specific exception type when checking
for RF-rack validation failures, so that other exception types can be
propagated properly.
In 8df61f6d99 we changed the requirements for creating materialized
views and MV-based indexes - instead of requiring the
rf_rack_valid_keyspaces flag to be set, we now require the keyspace to
be RF-rack-valid at the time of creation, and it is enforced to remain
RF-rack-valid while the MV exists. This validation is done in the cql
create view/index statements.
The same should be done also for alternator - when creating a table with
GSI or LSI, or when adding a GSI to an existing table, previously we
required the flag rf_rack_valid_keyspaces to be set. Now we change it to
instead check if the keyspace is RF-rack-valid, and if not the operation
fails with an appropriate error.
So far with oversampling the extended set of keys was returned from VS,
but query to the base table was still limited by the query `limit`.
Now for rescoring we want to fetch rows for all the keys returned from VS.
However later we need to restore the command limit, to trim result_set accordingly.
For non-rescoring scenarios we trim directly keys set returned from VS if it happens to exceed query limit.
With this change rescoring validation tests (except `no_nulls_in_rescored_results`) pass fully.
Fixes https://scylladb.atlassian.net/browse/SCYLLADB-83
This patch implements second part of rescoring - ordering results by similarity column added in earlier patch.
For this purpose in this patch we define `_ordering_comparator`, which enables pre-existing post-query ordering functionality.
However, none additional test passes yet, as they include ovesampling, which will be the subject of following patches.
Our plan for rescoring is to use the existing post-query ordering mechanism to sort (and trim) result_set by similarity column.
For general SELECT case this ordering is permitted only for queries with IN on the partition key and an ORDER BY, which is checked in `needs_post_query_ordering`.
Recently this check was overriden for ANN queries in https://github.com/scylladb/scylladb/pull/28109 to enable IN queries handled by VS without excessive post-processing.
In this patch we revert that change - ANN case will be handled by general check.
However we change the condition - we will enable post processing anytime `_ordering_comparator` is set.
In current implementation `_ordering_comparator` is created only in `select_statement::prepare` with `get_ordering_comparator`,
only for the same conditions as were checked in `needs_post_query_ordering`, so this change should be transparent for general SELECT.
For ANN query it is also not set (yet), so it will not influence ANN filtering, but we confirm that this functionality still works
by adding filtering test: `test/vector_search/filter_test.cc::vector_store_client_test_filtering_ann_cql`.
Rescoring ordering for ANN queries will be enabled when we add `_ordering_comparator` in following patch.
Rescoring consist of recalculating similarity score and reordering results based on it.
In this patch we add calculation of similarity score as a hidden (non-serialized) column and following patch will add reordering.
Normal ordering uses `add_column_for_post_processing`, however this works only for regular columns, not function.
So we create it together with user requested columns (this also forces the use of `selection_with_processing`) and hide the column later.
This also requires special handling for 'SELECT *' case - we need to manually add all columns before adding similarity column.
In case user already asks for similarity score in the SELECT clause, this value will be calculated twice - is should be optimized in future patches.
When setting up coredump handling, if there are old mounts in a deleted state (e.g. from an older installation),
systemd might fail to activate the new `.mount` unit properly because it assumes the path is already mounted.
Explicitly unmount `/var/lib/systemd/coredump` before proceeding with the setup to ensure a clean state.
Fix: scylladb/scylla-enterprise#5692Closesscylladb/scylladb#28300
Allows other topology operations to execute while tablets are being
drained on decommission. In particular, bootstrap on scale-out. This
is important for elasticity.
Allows multiple decommission/removenode to happen in parallel, which
is important for efficiency.
Flow of decommission/removenode request:
1) pending and paused, has tablet replicas on target node.
Tablet scheduler will start draining tablets.
2) No tablets on target node, request is pending but not paused
3) Request is scheduled, node is in transition
4) Request is done
Nodes are considered draining as soon as there is a leave or remove
request on them. If there are tablet replicas present on the target
node, the request is in a paused state and will not be picked by
topology coordinator. The paused state is computed from topology state
automatically on reload.
When request is not paused, its execution starts in
write_both_read_old state. The old tablet_draining state is not
entered (it's deprecated now).
Tablet load balancing will yield the state machine as soon as some
request is no longer paused and ready to be scheduled, based on
standard preemption mechanics.
Fixes#21452Closesscylladb/scylladb#24129
* https://github.com/scylladb/scylladb:
docs: Document parallel decommission and removenode and relevant task API
test: Add tests for parallel decommission/removenode
test: util: Introduce ensure_group0_leader_on()
test: tablets: Check that there are no migrations scheduled on draining nodes
test: lib: topology_builder: Introduce add_draining_request()
topology_coordinator, tablets: Fail draining operations when tablet migration fails due to critical disk utilization
tablets: topology_coordinator: Refactor to propagate reason for migration rollback
tablet_allocator: Skip co-location on draining nodes
node_ops: task_manager_module: Populate entity field also for active requests
tasks: node_ops: Put node id in the entity field
tasks, node_ops: Unify setting of task_stats in get_status() and get_stats()
topology: Protect against empty cancelation reason
tasks, topology: Make pending node operations abortable
doc: topology-over-raft.md: Fix diagram for replacing, tablet_draining is not engaged
raft_topology, tablets: Drain tablets in parallel with other topology operations
virtual_tables: Show draining and excluded fields in system.cluster_status and system.load_by_node
locator: topology: Add "draining" flag to a node
topology_coordinator: Extract generate_cancel_request_update()
storage_service: Drop dependency in topology_state_machine.hh in the header
locator: Extract common code in assert_rf_rack_valid_keyspace()
topology_coordinator, storage_service: Validate node removal/decommission at request submission time
When read or write operations are performed on a DC with RF=0 with LOCAL_QUORUM
or LOCAL_ONE consistency level, Cassandra throws `Unavailable` exception.
Scylla allowed such read operations and failed write operations with a cryptic:
"broken promise" error. This occured because the initial availability
check passed (quorum of 0 requires 0 replicas), but execution failed
later when no replicas existed to process the mutation.
This patch adds an explicit RF=0 validation for LOCAL_ONE and LOCAL_QUORUM that
throws before attempting operation execution.
The change also requires `test_query_dc_with_rf_0_does_not_crash_db` to be
upgraded. This testcase was asserting somewhat similar scenario, but wasn't
taking into account the whole matrix of combinations:
- scenarios: successful vs unsuccesful operation outcome
- local consistency levels: LOCAL_QUORUM & LOCAL_ONE
- operations: SELECT (read) & INSERT (write)
and so it's been extended to cover both the pre-existing and the current issues
and the whole matrix of combinations.
Fixes: scylladb/scylladb#27893
The core of `local_quorum_for()` has been extracted to
`get_replication_factor_for_dc()`, which is going to be used later,
while `local_quorum_for()` itself has been recreated using the exracted
part.
`network_topology_strategy` was abbreviated with `nrs`, and not `nts`. I
think someone incorrectly assumed it's 'network Replication strategy', hence
nrs.
The test is currently flaky. With `remove_dead_nodes_with == "remove"`,
it sends several ALTER KEYSPACE requests. The request performed just
after adding 3 new nodes can unexpectedly be sent twice to two
different nodes by the driver. The second receiver rejects the request
through the new guardrail added in 2e7ba1f8ce,
and the test fails.
This has been acknowledged as a bug in the Python driver. It shouldn't
retry non-idempotent requests with the default retry policy. There could
be one more bug in the driver, as it looks like the driver decides to
resend the request after it disconnects from the first receiver. The
first receiver has just bootstrapped, so the driver shouldn't disconnect.
We deflake the test by reconnecting the driver before performing the
problematic ALTER KEYSPACE request.
The change has been tested in byo, as the failure reproduces only in CI.
Without the change, the test fails once in ~250 runs in dev mode. With
the change, more than 1000 runs passed.
Fixes#27862
No backport needed as 2e7ba1f8ce is only
in master.
Closesscylladb/scylladb#28290
For the purpose of rescoring we will need information if the query is an ANN query
and the access to index option earlier in the `select_statement::prepare` than it happened before.
This patch refactors extracting this information to new helper structure `ann_ordering_info`
and is consistently using it.
Add enforce_rack_list option. When the option is set to true,
all tablet keyspaces have rack list replication factor.
When the option is on:
- CREATE STATEMENT always auto-extends rf to rack lists;
- ALTER STATEMENT fails when there is numeric rf in any DC.
The flag is set to false by default and a node needs to be restarted
in order to change its value. Starting a node with enforce_rack_list
option will fail, if there are any tablet keyspaces with numeric rf
in any DC.
enforce_rack_list is a per-node option and a user needs to ensure
that no tablet keyspace is altered or created while nodes in
the cluster don't have the consistent value.
Mark rf_rack_valid_keyspaces as deprecated.
Fixes: https://github.com/scylladb/scylladb/issues/26399.
New feature; no backport needed
Closesscylladb/scylladb#28084
* github.com:scylladb/scylladb:
test: add test for enforce_rack_list option
db: mark rf_rack_valid_keyspaces as deprecated
config: add enforce_rack_list option
Revert "alternator: require rf_rack_valid_keyspaces when creating index"
