This series introduces workload prioritization: an extension of the service levels feature which allows specifying "shares" per service level. The number of shares determines the priority of the user which has this service level attached (if multiple are attached then the one with the lowest shares wins).
Different service levels will be isolated in the following way:
- Each service level gets its own scheduling group with the number of shares (corresponding to the service level's number of shares), which controls the priority of the CPU and I/O used for user operations running on that service level.
- Each service level gets two reader concurrency semaphores, one for user reads and the other for read-before-write done for view updates.
- Each service level gets its own TCP connections for RPC to prevent priority inversion issues.
Because of the mandatory use of scheduling groups, which are a globally limited resource, the number of service levels is now limited to 7 user created service levels + 1 created by default that cannot be removed.
This feature has been previously only available in ScyllaDB Enterprise but has been made available for the source available ScyllaDB. The series was created by comparing the master branch with source-available-workbranch / enterprise branch and taking the workload prioritization related parts from the diff, then molding the resulting diff into a proper series. Some very minor changes were made such as fixing whitespace, removing unused or unnecessary code, adding some boilerplate (in api/) which was missing, but otherwise no major changes have been made.
No backport is required.
Closesscylladb/scylladb#22031
* github.com:scylladb/scylladb:
tracing: record scheduling group in trace event record
qos: un-shared-from-this standard_service_level_distributed_data_accessor
alternator: execute under scheduling group for service level
test.py: support multiple commands in prepare_cql in suite.yml
docs: add documentation for workload prioritization
docs/dev: describe workload prioritization features in service_levels
test/auth_cluster: test workload prioritization in service level tests
cqlpy/test_service_levels: add workload prioritization tests
api: introduce service levels specific API
api/cql_server_test: add information about scheduling group
db/virtual_tables: add scheduling group column to system.clients
test/boost: update service_level_controller_test for workload prio
qos: include number of shares in DESCRIBE
cql3/statements: update SL statements for workload prioritization
transport/server: use scheduling group assigned to current user
messaging_service: use separate set of connections per service levels
replica/database: add reader concurrency semaphore groups
qos: manage and assign scheduling groups to service levels
qos: use the shares field in service level reads/writes
qos: add shares to service_level_options
qos: explicitly specify columns when querying service level tables
db/system_distributed_keyspace: add shares column and upgrade code
db/system_keyspace: adjust SL schema for workload prioritization
gms: introduce WORKLOAD_PRIORITIZATION cluster feature
build: increase the max number of scheduling groups
qos: return correct error code when SL does not exist
This patch sets up an `alien_worker`, `advanced_rpc_compression::tracker`,
`dict_sampler` and `dictionary_service` in `main()`, and wires them to each other
and to `messaging_service`.
`messaging_service` compresses its network traffic with compressors managed by
the `advanced_rpc_compression::tracker`. All this traffic is passed as a single
merged "stream" through `dict_sampler`.
`dictionary_service` has access to `dict_sampler`.
On chosen nodes (by default: the Raft leader), it uses the sampler to maintain
a random multi-megabyte sample of the sampler's stream. Every several minutes,
it copies the sample, trains a compression dictionary on it (by calling zstd's
training library via the `alien_worker` thread) and publishes the new dictionary
to `system.dicts` via Raft.
This update triggers a callback into `advanced_rpc_compression::tracker` on all nodes,
which updates the dictionary used by the compressors it manages.
Adds glue needed to pass lz4 and zstd with streaming and/or dictionaries
as the network traffic compressors for Seastar's RPC servers.
The main jobs of this glue are:
1. Implementing the API expected by Seastar from RPC compressors.
2. Expose metrics about the effectiveness of the compression.
3. Allow dynamically switching algorithms and dictionaries on a running
connection, without any extra waits.
The biggest design decision here is that the choice of algorithm and dictionary
is negotiated by both sides of the connection, not dictated unilaterally by the
sender.
The negotiation algorithm is fairly complicated (a TLA+ model validating
it is included in the commit). Unilateral compression choice would be much simpler.
However, negotiation avoids re-sending the same dictionary over every
connection in the cluster after dictionary updates (with one-way communication,
it's the only reliable way to ensure that our receiver possesses the dictionary
we are about to start using), lets receivers ask for a cheaper compression mode
if they want, and lets them refuse to update a dictionary if they don't think
they have enough free memory for that.
In hindsight, those properties probably weren't worth the extra complexity and
extra development effort.
Zstd can be quite expensive, so this patch also includes a mechanism which
temporarily downgrades the compressor from zstd to lz4 if zstd has been
using too much CPU in a given slice of time. But it should be noted that
this can't be treated as a reliable "protection" from negative performance
effects of zstd, since a downgrade can happen on the sender side,
and receivers are at the mercy of senders.
Currently truncating a table works by issuing an RPC to all the nodes which call `database::truncate_table_on_all_shards()`, which makes sure that older writes are dropped.
It works with tablets, but is not safe. A concurrent replication process may bring back old data.
This change makes makes TRUNCATE TABLE a topology operation, so that it excludes with other processes in the system which could interfere with it. More specifically, it makes TRUNCATE a global topology request.
Backporting is not needed.
Fixes#16411Closesscylladb/scylladb#19789
* github.com:scylladb/scylladb:
docs: docs: topology-over-raft: Document truncate_table request
storage_proxy: fix indentation and remove empty catch/rethrow
test: add tests for truncate with tablets
storage_proxy: use new TRUNCATE for tablets
truncate: make TRUNCATE a global topology operation
storage_service: move logic of wait_for_topology_request_completion()
RPC: add truncate_with_tablets RPC with frozen_topology_guard
feature_service: added cluster feature for system.topology schema change
system.topology_requests: change schema
storage_proxy: propagate group0 client and TSM dependency
The goal of merge is to reduce the tablet count for a shrinking table. Similar to how split increases the count while the table is growing. The load balancer decision to merge is implemented today (came with infrastructure introduced for split), but it wasn't handled until now.
Initial tablet count is respected while the table is in "growing mode". For example, the table leaves it if there was a need to split above the initial tablet count. After the table leaves the mode, the average size can be trusted to determine that the table is shrinking. Merge decision is emitted if the average tablet size is 50% of the target. Hysteresis is applied to avoid oscillations between split and merges.
Similar to split, the decision to merge is recorded in tablet map's resize_type field with the string "merge". This is important in case of coordinator failover, so new coordinator continues from where the old left off.
Unlike split, the preparation phase during merge is not done by the replica (with split compactions), but rather by the coordinator by co-locating sibling tablets in the same node's shard. We can define sibling tablets as tablets that have contiguous range and will become one after merge. The concept is based on the power-of-two constraint and token contiguity. For example, in a table with 4 tablets, tablets of ids 0 and 1 are siblings, 2 and 3 are also siblings.
The algorithm for co-locating sibling tablets is very simple. The balancer is responsible for it, and it will emit migrations so that "odd" tablet will follow the "even" one. For example, tablet 1 will be migrated to where tablet 0 lives. Co-location is low in priority, it's not the end of the world to delay merge, but it's not ideal to delay e.g. decommission or even regular load balancing as that can translate into temporary unbalancing, impacting the user activities. So co-location migrations will happen when there is no more important work to do.
While regular balancing is higher in priority, it will not undo the co-location work done so far. It does that by treating co-located tablets as if they were already merged. The load inversion convergence check was adjusted so balancer understand when two tablets are being migrated instead of one, to avoid oscillations.
When balancer completes co-location work for a table undergoing merge, it will put the id of the table into the resize_plan, which is about communicating with the topology coordinator that a table is ready for it. With all sibling tablets co-located, the coordinator can resize the tablet map (reduce it by a factor of 2) and record the new map into group0. All the replicas will react to it (on token metadata update) by merging the storage (memtable(s) + sstables) of sibling tablets into one.
Fixes#18181.
system test details:
test: https://github.com/pehala/scylla-cluster-tests/blob/tablets_split_merge/tablets_split_merge_test.py
yaml file: https://github.com/pehala/scylla-cluster-tests/blob/tablets_split_merge/test-cases/features/tablets/tablets-split-merge-test.yaml
instance type: i3.8xlarge
nodes: 3
target tablet size: 0.5G (scaled down by 10, to make it easier to trigger splits and merges)
description: multiple cycles of growing and shrinking the data set in order to trigger splits and merges.
data_set_size: ~100G
initial_tablets: 64, so it grew to 128 tablets on split, and back to 64 on merge.
latency of reads and writes that happened in parallel to split and merge:
```
$ for i in scylla-bench*; do cat $i | grep "Mode\|99th:\|99\.9th:"; done
Mode: write
99.9th: 3.145727ms
99th: 1.998847ms
99.9th: 3.145727ms
99th: 2.031615ms
Mode: read
99.9th: 3.145727ms
99th: 2.031615ms
99.9th: 3.145727ms
99th: 2.031615ms
Mode: write
99.9th: 3.047423ms
99th: 1.933311ms
99.9th: 3.047423ms
99th: 1.933311ms
Mode: read
99.9th: 3.145727ms
99th: 1.900543ms
99.9th: 3.145727ms
99th: 1.900543ms
Mode: write
99.9th: 5.079039ms
99th: 3.604479ms
99.9th: 35.389439ms
99th: 25.624575ms
Mode: write
99.9th: 3.047423ms
99th: 1.998847ms
99.9th: 3.047423ms
99th: 1.998847ms
Mode: read
99.9th: 3.080191ms
99th: 2.031615ms
99.9th: 3.112959ms
99th: 2.031615ms
```
Closesscylladb/scylladb#20572
* github.com:scylladb/scylladb:
docs: Document tablet merging
tests/boost: Add test to verify correctness of balancer decisions during merge
tests/topology_experimental_raft: Add tablet merge test
service: Handle exception when retrying split
service: Co-locate sibling tablets for a table undergoing merge
gms: Add cluster feature for tablet merge
service: Make merge of resize plan commutative
replica: Implement merging of compaction groups on merge completion
replica: Handle tablet merge completion
service: Implement tablet map resize for merge
locator: Introduce merge_tablet_info()
service: Rename topology::transition_state::tablet_split_finalization
service: Respect initial_tablet_count if table is in growing mode
service: Wire migration_tablet_set into the load balancer
locator: Add tablet_map::sibling_tablets()
service: Introduce sorted_replicas_for_tablet_load()
locator/tablets: Extend tablet_replica equality comparator to three-way
service: Introduce alias to per-table candidate map type
service: Add replication constraint check variant for migration_tablet_set
service: Add convergence check variant for migration_tablet_set
service: Add migration helpers for migration_tablet_set
service/tablet_allocator: Introduce migration_tablet_set
service: Introduce migration_plan::add(migrations_vector)
locator/tablets: Introduce tablet_map::for_each_sibling_tablets()
locator/tablets: Introduce tablet_map::needs_merge()
locator/tablets: Introduce resize_decision::initial_decision()
locator/tablets: Fix return type of three-way comparison operators
service: Extract update of node load on migrations
service: Extract converge check for intra-node migration
service: Extract erase of tablet replicas from candidate list
scripts/tablet-mon: Allow visualization of tablet id
Small adjustments and improvements to the documentation in the raft
section.
Fixing Markdown lint warnings:
- MD004/ul-style: Unordered list style [Expected: dash; Actual: asterisk]
- MD007/ul-indent: Unordered list indentation [Expected: 0; Actual: 2]
- MD032/blanks-around-lists: Lists should be surrounded by blank lines
- MD036/no-emphasis-as-heading: Emphasis used instead of a heading
- MD046/code-block-style: Code block style [Expected: fenced; Actual: indented]
Closesscylladb/scylladb#21780
This transition state will be reused by merge completion, so let's
rename it to tablet_resize_finalization.
The completion handling path will also be reused, so let's rename
functions involved similarly.
The old name "tablet split finalization" is deprecated but still
recognized and points to the correct transition. Otherwise, the
reverse lookup would fail when populating topology system table
which last state was split finalization.
NOTE:
I thought of adding a new tablet_merge_finalization, but it would
complicate things since more than one table could be ready for
either split or merge, so you need a generic transition state
for handling resize completion.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Schema of system tables is defined statically and table_schema_version needs to be explicitly set in code like this:
```
builder.with_version(system_keyspace::generate_schema_version(table_id, version_offset));
```
Whenever schema is changed, the schema version needs to change, otherwise we hit undefined behavior when trying to interpret mutation data created with the old schema using the new schema.
It's not obvious that one needs to do that and developers often forget to do that. There were several instances of mistakes of omission, some caught during review, some not, e.g.: 31ea74b96e.
This patch changes definitions to call the new `schema_builder::with_hash_version()`, which will make the schema builder compute version from schema definition so that changes of the schema will automatically change the version. This way we no longer rely on the developer to remember to bump the version offset.
All nodes should arrive at the same version, which is verified by existing `test_group0_schema_versioning` and a new unit test: `test_system_schema_version_is_stable`.
Closesscylladb/scylladb#21602
* github.com:scylladb/scylladb:
system_tables: Compute schema version automatically
schema_builder: Introduce with_hash_version()
schema: Store raw_view_info in schema::raw_schema
schema: Remove dead comment
hashing: Add hasher for unordered_map
hashing: Add hasher for unique_ptr
hashing: Add hasher for double
[avi: add missing include <memory> to hashing.hh]
This depends on the previous change to the schema_builder
which makes version computation depend on definition only
instead of being new time uuid.
This way we avoid the possibility for a common mistake
when schema of a system table is extended but we forget
to bump up its version passed to .with_version().
Python and Python developers don't like directory names to include a
minus sign, like "cql-pytest". In this patch we rename test/cql-pytest
to test/cqlpy, and also change a few references in other code (e.g., code
that used test/cql-pytest/run.py) and also references to this test suite
in documentation and comments.
Arguably, the word "test" was always redundant in test/cql-pytest, and
I want to leave the "py" in test/cqlpy to emphasize that it's Python-based
tests, contrasting with test/cql which are CQL-request-only approval
tests.
Fixes#20846
Signed-off-by: Nadav Har'El <nyh@scylladb.com>
Cassandra 4.1 announced a new option to create a role with:
`HASHED PASSWORD`. Example:
```
CREATE ROLE bob WITH HASHED PASSWORD = 'hashed_password';
```
We've already introduced another option following the same
semantics: `SALTED HASH`; example:
```
CREATE ROLE bob WITH SALTED HASH = 'salted_hash';
```
The change hasn't made it to any release yet, so in this commit
we rename it to `HASHED PASSWORD` to be compatible with Cassandra.
Additionally, we adjust existing tests to work against Cassandra too.
Fixesscylladb/scylladb#21350Closesscylladb/scylladb#21352
Single-row reads from large partition issue 64 KiB reads to the data file,
which is equal to the default span of the promoted index block in the data file.
If users would want to increase selectivity of the index to speed up single-row reads,
this won't be effective. The reason is that the reader uses promoted index
to look up the start position in the data file of the read, but end position
will in practice extend to the next partition, and amount of I/O will be
determined by the underlying file input stream implementation and its
read-ahead heuristics. By default, that results in at least 2 IOs 32KB each.
There is already infrastructure to lookup end position based on upper
bound of the read, in anticipation for sharing the promoted index cache,
but it's not effective becasue it's a non-populating lookup and the upper
bound cursor has its own private cached_promoted_index, which is cold
when positions are computed. It's non-populating on purpose, to avoid
extra index file IO to read upper bound. In case upper bound is far-enough
from the lower bound, this will only increase the cost of the read.
The solution employed here is to warm up the lower bound cursor's
cache before positions are computed, and use that cursor for
non-populating lookup of the upper bound.
We use the lower bound cursor and the slice's lower bound so that we
read the same blocks as later lower-bound slicing would, so that we
don't incur extra IO for cases where looking up upper bound is not
worth it, that is when upper bound is far from the lower bound. If
upper bound is near lower bound, then warming up using lower bound
will populate cached_promoted_index with blocks which will allow us to
locate the upper bound block accurately. This is especially important
for single-row reads, where the bounds are around the same key. In
this case we want to read the data file range which belongs to a
single promoted index block. It doesn't matter that the upper bound
is not exactly the same. They both will likely lie in the same block,
and if not, binary search will bring adjacent blocks into cache. Even
if upper bound is not near, the binary search will populate the cache
with blocks which can be used to narrow down the data file range
somewhat.
Fixes#10030.
The change was tested with perf-fast-forward.
I populated the data set with `column_index_size_in_kb` set to 1
scylla perf-fast-forward --populate --run-tests=large-partition-slicing --column-index-size-in-kb=1
Test run:
build/release/scylla perf-fast-forward --run-tests=large-partition-select-few-rows -c1 --keep-cache-across-test-cases --test-case-duration=0
This test issues two reads of subsequent keys from the middle of a large partition (1M rows in total). The first read will miss in the index file page cache, the second read will hit.
Notice that before the change, the second read issued 2 aio requests worth of 64KiB in total.
After the change, the second read issued 1 aio worth of 2 KiB. That's because promoted index block is larger than 1 KiB.
I verified using logging that the data file range matches a single promoted index block.
Also, the first read which misses in cache is still faster after the change.
Before:
```
running: large-partition-select-few-rows on dataset large-part-ds1
Testing selecting few rows from a large partition:
stride rows time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu
500000 1 0.009802 1 1 102 0 102 102 21.0 21 196 2 1 0 1 1 0 0 0 568 269 4716050 53.4%
500001 1 0.000321 1 1 3113 0 3113 3113 2.0 2 64 1 0 1 0 0 0 0 0 116 26 555110 45.0%
```
After:
```
running: large-partition-select-few-rows on dataset large-part-ds1
Testing selecting few rows from a large partition:
stride rows time (s) iterations frags frag/s mad f/s max f/s min f/s avg aio aio (KiB) blocked dropped idx hit idx miss idx blk c hit c miss c blk allocs tasks insns/f cpu
500000 1 0.009609 1 1 104 0 104 104 20.0 20 137 2 1 0 1 1 0 0 0 561 268 4633407 43.1%
500001 1 0.000217 1 1 4602 0 4602 4602 1.0 1 2 1 0 1 0 0 0 0 0 110 26 313882 64.1%
```
Backports: none, not a regression
Closesscylladb/scylladb#20522
* github.com:scylladb/scylladb:
perf: perf_fast_forward: Add test case for querying missing rows
perf-fast-forward: Allow overriding promoted index block size
perf-fast-forward: Test subsequent key reads from the middle in test_large_partition_select_few_rows
perf-fast-forward: Allow adding key offset in test_large_partition_select_few_rows
perf-fast-forward: Use single-partition reads in test_large_partition_select_few_rows
sstables: bsearch_clustered_cursor: Add more tracing points
sstables: reader: Log data file range
sstables: bsearch_clustered_cursor: Unify skip_info logging
sstables: bsearch_clustered_cursor: Narrow down range using "end" position of the block
sstables: bsearch_clustered_cursor: Skip even to the first block
test: sstables: sstable_3_x_test: Improve failure message
sstables: mx: writer: Never include partition_end marker in promoted index block width
sstables: Reduce amount of I/O for clustering-key-bounded reads from large partitions
sstables: clustered_cursor: Track current block
Commit 3a12ad96c7
added an sstable_identifier uuid to the SSTable
scylla_metadata component, however it was
under-documented and this patch adds the missing
documentation for the sstable component format,
and to the scylla sstable tool documentation.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Closesscylladb/scylladb#21221
this commit moves the object storage configuration guide from the developer
documentation to the user-facing admin documentation. the change reflects
the increasing importance of object storage integration in user-facing
features.
in this change:
- move relevant content from `docs/dev/object_storage.md` to
`docs/operating-scylla/admin.rst`
- reformat the content from Markdown to reStructuredText (RST)
- reword and restructure the content to be more user-friendly
- add explanations and context suitable for a broader audience
this change makes the object storage configuration information more
accessible to Scylla administrators and end-users, supporting the adoption
of new features built on top of object storage integration.
Signed-off-by: Kefu Chai <kefu.chai@scylladb.com>
Currently, it may happen that the last promoted index block includes
the partition_end marker. That's because we first write the partition
end marker and then emit the unclosed block. This behavior matches
Cassandra (checked in 3.x and 5.0.1).
This is problematic for ruling out data file reads based on index.
The width field is currently unused, but it will be used later where
the width of the last block is used to compute the skip position past
the last block for lookups which land after all keys in the
partition. If width includes the marker then such a skip would land in
the next partition, which is incorrect, as the reader context expects
a cell element. Even if that was recognized, it's wrong - if this is
not a single partition read (so upper bound is not at the next
partition too), then we would read from the wrong (next) partition.
We want to be able to make such skips in order to avoid unnecessary
data file IO for reads of missing rows. Currently, we would always
read the last block even if the key is past its "end" position.
Another way to solve this would be to propagate the "past the last
block" condition from the index cursor to the reader and let it deal
with it, but the logic for that would be complicated. With this fix,
there is no special logic required.
with this parameter, "backup" API can backup the given table, this
enables it to be a drop-in replacement of existing rclone API used by
scylla manager.
Fixes https://github.com/scylladb/scylladb/issues/20636
---
this change is a part of the efforts to bring the native backup/restore to scylla, no need to backprt.
Closesscylladb/scylladb#20661
* github.com:scylladb/scylladb:
backup_task: fix the indent
treewide: add "table" parameter to "backup" API
with this parameter, "backup" API can backup the given table, this
enables it to be a drop-in replacement of existing rclone API used by
scylla manager.
in this change:
* api/storage_service: add "table" parameter to "backup" API.
* snapshot_ctl: compose the full path of the snapshot directory in
`snapshot_ctl::start_backup`. since we have all the information
for composing the snapshot directory, and what the `backup_task_impl`
class is interested is but the snapshot directory, we just pass
the path to it instead the individual components of the directory.
* backup_task_impl: instead of scan the whole keyspace recursively,
only scan the specified snapshot directory.
Fixesscylladb/scylladb#20636
Signed-off-by: Kefu Chai <kefu.chai@scylladb.com>
Auth has been managed via Raft since Scylla 6.0. Restoring data
following the usual procedure (1) is error-prone and so a safer
method must have been designed and implemented. That's what
happens in this PR.
We want to extend `DESC SCHEMA` by auth and service levels
to provide a safe way to backup and restore those two components.
To realize that, we change the meaning of `DESC SCHEMA WITH INTERNALS`
and add a new "tier": `DESC SCHEMA WITH INTERNALS AND PASSWORDS`.
* `DESC SCHEMA` -- no change, i.e. the statement describes the current
schema items such as keyspaces, tables, views, UDTs, etc.
* `DESC SCHEMA WITH INTERNALS` -- does the same as the previous tier
and also describes auth and service levels. No information about
passwords is returned.
* `DESC SCHEMA WITH INTERNALS AND PASSWORDS` -- does the same
as the previous tier and also includes information about the salted
hashes corresponding to the passwords of roles.
To restore existing roles, we extend the `CREATE ROLE` statement
by allowing to use the option `WITH SALTED HASH = '[...]'`.
---
Implementation strategy:
* Add missing things/adjust existing ones that will be used later.
* Implement creating a role with salted hash.
* Add tests for creating a role with salted hash.
* Prepare for implementing describe functionality of auth and service levels.
* Implement describe functionality for elements of auth and service levels.
* Extend the grammar.
* Add tests for describe auth and service levels.
* Add/update documentation.
---
(1): https://opensource.docs.scylladb.com/stable/operating-scylla/procedures/backup-restore/restore.html
In case the link stops working, restoring a schema was realised
by managing raw files on disk.
Fixesscylladb/scylladb#18750Fixesscylladb/scylladb#18751Fixesscylladb/scylladb#20711Closesscylladb/scylladb#20168
* github.com:scylladb/scylladb:
docs: Update user documentation for backup and restore
docs/dev: Add documentation for DESC SCHEMA
test: Add tests for describing auth and service levels
cql3/functions/user_function: Remove newline character before and after UDF body
cql3: Implement DESCRIBE SCHEMA WITH INTERNALS AND PASSWORDS
auth: Implement describing auth
auth/authenticator: Add member functions for querying password hash
service/qos/service_level_controller: Describe service levels
data_dictionary: Remove keyspace_element.hh
treewide: Start using new overloads of describe
treewide: Fix indentation in describe functions
treewide: Return create statement optionally in describe functions
treewide: Add new describe overloads to implementations of data_dictionary::keyspace_element
treewide: Start using schema::ks_name() instead of schema::keyspace_name()
cql3: Refactor `description`
cql3: Move description to dedicated files
test: Add tests for `CREATE ROLE WITH SALTED HASH`
cql3/statements: Restrict CREATE ROLE WITH SALTED HASH
auth: Allow for creating roles with SALTED HASH
types: Introduce a function `cql3_type_name_without_frozen()`
cql3/util: Accept std::string_view rather than const sstring&
We add documentation for developers addressing
`DESCRIBE SCHEMA`. It covers the following aspects
of it:
* motivation,
* synopsis of the solution,
* implementation of the solution,
as well as a few subsections explaining the details:
* restoring process and its side effects,
* restoring roles with passwords,
* list of statements generated by `DESC SCHEMA`
with examples,
* implementation details.
`unspecified` workload type is an internal value and it's not exposed to
user via CQL.
Default value for workload type from user's perspective is `NULL`.
Fixesscylladb/scylladb#20780
* Also dump diagnostics when a read times out while active (not queued).
* Add the "Trigger permit" line, containing the details of the permit which caused the diagnostics dump (by e.g. timing out).
* Add the "Identified bottleneck(s)" line, containing the identified bottlenecks which lead to permits being queued. This line is missing if no such bottleneck can be identified.
* Document the new features, as well as the stat dump, which was added some time ago.
Example of the new dump format:
```
INFO 2024-09-12 08:09:48,046 [shard 0:main] reader_concurrency_semaphore - Semaphore reader_concurrency_semaphore_dump_reader_diganostics with 8/10 count and 106192275/32768 memory resources: timed out, dumping permit diagnostics:
Trigger permit: count=0, memory=0, table=ks.tbl0, operation=mutation-query, state=waiting_for_admission
Identified bottleneck(s): memory
permits count memory table/operation/state
3 2 26M *.*/push-view-updates-2/active
3 2 16M ks.tbl1/push-view-updates-1/active
1 1 15M ks.tbl2/push-view-updates-1/active
1 0 13M ks.tbl1/multishard-mutation-query/active
1 0 12M ks.tbl0/push-view-updates-1/active
1 1 10M ks.tbl3/push-view-updates-2/active
1 1 6060K ks.tbl3/multishard-mutation-query/active
2 1 1930K ks.tbl0/push-view-updates-2/active
1 0 1216K ks.tbl0/multishard-mutation-query/active
6 0 0B ks.tbl1/shard-reader/waiting_for_admission
3 0 0B *.*/data-query/waiting_for_admission
9 0 0B ks.tbl0/mutation-query/waiting_for_admission
2 0 0B ks.tbl2/shard-reader/waiting_for_admission
4 0 0B ks.tbl0/shard-reader/waiting_for_admission
9 0 0B ks.tbl0/data-query/waiting_for_admission
7 0 0B ks.tbl3/mutation-query/waiting_for_admission
5 0 0B ks.tbl1/mutation-query/waiting_for_admission
2 0 0B ks.tbl2/mutation-query/waiting_for_admission
8 0 0B ks.tbl1/data-query/waiting_for_admission
1 0 0B *.*/mutation-query/waiting_for_admission
26 0 0B permits omitted for brevity
96 8 101M total
Stats:
permit_based_evictions: 0
time_based_evictions: 0
inactive_reads: 0
total_successful_reads: 0
total_failed_reads: 0
total_reads_shed_due_to_overload: 0
total_reads_killed_due_to_kill_limit: 0
reads_admitted: 1
reads_enqueued_for_admission: 82
reads_enqueued_for_memory: 0
reads_admitted_immediately: 1
reads_queued_because_ready_list: 0
reads_queued_because_need_cpu_permits: 82
reads_queued_because_memory_resources: 0
reads_queued_because_count_resources: 0
reads_queued_with_eviction: 0
total_permits: 97
current_permits: 96
need_cpu_permits: 0
awaits_permits: 0
disk_reads: 0
sstables_read: 0
```
Fixes: https://github.com/scylladb/scylladb/issues/19535
Improvement, no backport needed.
Closesscylladb/scylladb#20545
* github.com:scylladb/scylladb:
docs/dev/reader-concurrency-semaphore.md: update the documentation on diagnostics dumps
test/boost/reader_concurrency_semaphore_test: test the new diagnostics functionality
reader_concurrency_semaphore: add bottleneck self-diagnosis to diagnosis dump
reader_concurrency_semaphore: include trigger permit in diagnostic dump
reader_concurrency_semaphore: propagate permit to do_dump_reader_permit_diagnostics()
reader_concurrency_semaphore: use consistent exception type for timeout
reader_concurrency_semaphore: dump diagnostics when non-waiting reader times out
When purging regular tombstone consult the min_live_timestamp, if available.
This is safe since we don't need to protect dead data from resurrection, as it is already dead.
For shadowable_tombstones, consult the min_memtable_live_row_marker_timestamp,
if available, otherwise fallback to the min_live_timestamp.
If we see in a view table a shadowable tombstone with time T, then in any row where the row marker's timestamp is higher than T the shadowable tombstone is completely ignored and it doesn't hide any data in any column, so the shadowable tombstone can be safely purged without any effect or risk resurrecting any deleted data.
In other words, rows which might cause problems for purging a shadowable tombstone with time T are rows with row markers older or equal T. So to know if a whole sstable can cause problems for shadowable tombstone of time T, we need to check if the sstable's oldest row marker (and not oldest column) is older or equal T. And the same check applies similarly to the memtable.
If both extended timestamp statistics are missing, fallback to the legacy (and inaccurate) min_timestamp.
Fixesscylladb/scylladb#20423Fixesscylladb/scylladb#20424
> [!NOTE]
> no backport needed at this time
> We may consider backport later on after given some soak time in master/enterprise
> since we do see tombstone accumulation in the field under some materialized views workloads
Closesscylladb/scylladb#20446
* github.com:scylladb/scylladb:
cql-pytest: add test_compaction_tombstone_gc
sstable_compaction_test: add mv_tombstone_purge_test
sstable_compaction_test: tombstone_purge_test: test that old deleted data do not inhibit tombstone garbage collection
sstable_compaction_test: tombstone_purge_test: add testlog debugging
sstable_compaction_test: tombstone_purge_test: make_expiring: use next_timestamp
sstable, compaction: add debug logging for extended min timestamp stats
compaction: get_max_purgeable_timestamp: use memtable and sstable extended timestamp stats
compaction: define max_purgeable_fn
tombstone: can_gc_fn: move declaration to compaction_garbage_collector.hh
sstables: scylla_metadata: add ext_timestamp_stats
compaction_group, storage_group, table_state: add extended timestamp stats getters
sstables, memtable: track live timestamps
memtable_encoding_stats_collector: update row_marker: do nothing if missing
The part of the document which explains diagnostics dumps was due for an
update. It was missing an explanation on the dumped stats and it
also needs to explain the "Problematic permit" and "Identified
bottleneck(s)".
Store and retrieve the optional extended timestamp statistics
(min_live_timestamp and min_live_row_marker_timestamp)
in the scylla_metadata component.
Note that there is no need for a cluster feature to
store those attributes since the scylla_metadata
on-disk format is extensible so that old sstables
can be read by new versions, seeing the extra stats
is missing, and new sstables can be read by old
versions that ignore unknown scylla metadata section types.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Refs #18161
Yet another approach to dealing with large commitlog submissions.
We handle oversize single mutation by adding yet another entry
typo: fragmented. In this case we only add a fragment (aha) of
the data that needs storing into each entry, along with metadata
to correlate and reconstruct the full entry on replay.
Because these fragmented entries are spread over N segments, we
also need to add references from the first segment in a chain
to the subsequent ones. These are released once we clear the
relevant cf_id count in the base.
*
This approach has the downside that due to how serialization etc
works w.r.t. mutations, we need to create an intermediate buffer
to hold the full serialized target entry. This is then incrementally
written into entries of < max_mutation_size, successively requesting
more segments.
On replay, when encountering a fragment chain, the fragment is
added to a "state", i.e. a mapping of currently processing
frag chains. Once we've found all fragments and concatenated
the buffers into a single fragmented one, we can issue a
replay callback as usual.
Note that a replay caller will need to create and provide such
a state object. Old signature replay function remains for tests
and such.
This approach bumps the file format (docs to come).
To ensure "atomicity" we both force synchronization, and should
the whole op fail, we restore segment state (rewinding), thus
discarding data all we wrote.
Closesscylladb/scylladb#19472
* github.com:scylladb/scylladb:
commitlog/database: Make some commitlog options updatable + add feature listener
features/config: Add feature for fragmented commitlog entries
docs: Add entry on commitlog file format v4
commitlog_test: Add more oversized cases
commitlog_replayer: Replay segments in order created
commitlog_replayer: Use replay state to support fragmented entries
commitlog_replayer: coroutinize partly
commitlog: Handle oversized entries
~~~
What we have today in "docs/dev/docker-hub.md" on "aio-max-nr" dates back
to scylla commit f4412029f4 ("docs/docker-hub.md: add quickstart section
with --smp 1", 2020-09-22). Problems with the current language:
- The "65K" claim as default value on non-production systems is wrong;
"fs/aio.c" in Linux initializes "aio_max_nr" to 0x10000, which is 64K.
- The section in question uses equal signs (=) incorrectly. The intent was
probably to say "which means the same as", but that's not what equality
means.
- In the same section, the relational operator "<" is bogus. The available
AIO count must be at least as high (>=) as the requested AIO count.
- Clearer names should be used;
adjust_max_networking_aio_io_control_blocks() in "src/core/reactor.cc"
sets a great example:
- "reactor::max_aio" should be called "storage_iocbs",
- "detect_aio_poll" should be called "preempt_iocbs",
- "reactor_backend_aio::max_polls" should be called "network_iocbs".
- The specific value 10000 for the last one ("network_iocbs") is not
correct in scylla's context. It is correct as the Seastar default, but
scylla has used 50000 since commit 2cfc517874 ("main, test: adjust
number of networking iocbs", 2021-07-18).
Rewrite the section to address these problems.
See also:
- https://github.com/scylladb/scylladb/issues/5981
- https://github.com/scylladb/seastar/pull/2396
- https://github.com/scylladb/scylladb/pull/19921
Signed-off-by: Laszlo Ersek <laszlo.ersek@scylladb.com>
~~~
No need for backporting; the documentation being refreshed targets developers as audience, not end-users.
Closesscylladb/scylladb#20398
* github.com:scylladb/scylladb:
docs/dev/docker-hub.md: refresh aio-max-nr calculation
docs/dev/docker-hub.md: strip trailing whitespace
Before the introduction of "scripts/refresh-submodules.sh", there was
indeed some manual work for the maintainer to do, hence "publish your
work" must have sounded correct. Today, the phrase "publish your work"
sounds confusing.
Commit 71da4e6e79 ("docs: Document sync-submodules.sh script in
maintainer.md", 2020-06-18) should have arguably reworded the last step of
the submodule refresh procedure; let's do it now.
Signed-off-by: Laszlo Ersek <laszlo.ersek@scylladb.com>
Closesscylladb/scylladb#20333
What we have today in "docs/dev/docker-hub.md" on "aio-max-nr" dates back
to scylla commit f4412029f4 ("docs/docker-hub.md: add quickstart section
with --smp 1", 2020-09-22). Problems with the current language:
- The "65K" claim as default value on non-production systems is wrong;
"fs/aio.c" in Linux initializes "aio_max_nr" to 0x10000, which is 64K.
- The section in question uses equal signs (=) incorrectly. The intent was
probably to say "which means the same as", but that's not what equality
means.
- In the same section, the relational operator "<" is bogus. The available
AIO count must be at least as high (>=) as the requested AIO count.
- Clearer names should be used;
adjust_max_networking_aio_io_control_blocks() in "src/core/reactor.cc"
sets a great example:
- "reactor::max_aio" should be called "storage_iocbs",
- "detect_aio_poll" should be called "preempt_iocbs",
- "reactor_backend_aio::max_polls" should be called "network_iocbs".
- The specific value 10000 for the last one ("network_iocbs") is not
correct in scylla's context. It is correct as the Seastar default, but
scylla has used 50000 since commit 2cfc517874 ("main, test: adjust
number of networking iocbs", 2021-07-18).
Rewrite the section to address these problems.
See also:
- https://github.com/scylladb/scylladb/issues/5981
- https://github.com/scylladb/seastar/pull/2396
- https://github.com/scylladb/scylladb/pull/19921
Signed-off-by: Laszlo Ersek <laszlo.ersek@scylladb.com>
* in the "Backporting Seastar commits" section, there's a single quote
instead of a backtick in this line, so fix it.
* add backticks around `refresh-submodules.sh`, which is a filename.
* correct the command line setting a git config option, because `git-config`
does not support this command line syntax,
```console
$ git config --global diff.conflictstyle = diff3
$ git config --global get diff.conflictstyle
=
$ git config --global diff.conflictstyle diff3
$ git config --global get diff.conflictstyle
diff3
```
quote from git-config(1)
> ```
> git config set [<file-option>] [--type=<type>] [--all] [--value=<value>] [--fixed-value] <name> <value>
> ```
* stop using the deprecated mode of the `git-config` command, and use
subcommand instead. as git-config(1) puts:
> git config <name> <value> [<value-pattern>]
> Replaced by git config set [--value=<pattern>] <name> <value>.
Signed-off-by: Kefu Chai <kefu.chai@scylladb.com>
Closesscylladb/scylladb#20328
Currently the doc assumes that object storage can only be used to keep
sstables on it. It's going to change, restructure the doc to allow for
more usage scenarios.
