Add ability to skip read-only replicas for INSERT into Distributed engine

[azat]

Changelog category (leave one):

• Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Add ability to skip read-only replicas for INSERT into Distributed engine (Controlled with distributed_insert_skip_read_only_replicas setting, by default OFF - backward compatible)

Sometimes replica may be readonly for a long time, and this will make
some INSERT queries fail, but it does not make sense to INSERT into
readonly replica, so let's ignore them.

But note, that this will require to extend TableStatus (not extend, but
introduce new version), that will have is_readonly field.

Also before background INSERT into Distributed does not uses
getManyChecked() which means that they do not request TableStatus
packet, while now they would, though this is minor (just a note).

[robot-clickhouse-ci-1]

This is an automated comment for commit 4eafb86 with description of existing statuses. It's updated for the latest CI running

❌ Click here to open a full report in a separate page

Check name	Description	Status
CI running	A meta-check that indicates the running CI. Normally, it's in success or pending state. The failed status indicates some problems with the PR	⏳ pending
Performance Comparison	Measure changes in query performance. The performance test report is described in detail here. In square brackets are the optional part/total tests	❌ failure

Successful checks

Check name	Description	Status
A Sync	There's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS	✅ success
AST fuzzer	Runs randomly generated queries to catch program errors. The build type is optionally given in parenthesis. If it fails, ask a maintainer for help	✅ success
ClickBench	Runs [ClickBench](https://github.com/ClickHouse/ClickBench/) with instant-attach table	✅ success
ClickHouse build check	Builds ClickHouse in various configurations for use in further steps. You have to fix the builds that fail. Build logs often has enough information to fix the error, but you might have to reproduce the failure locally. The cmake options can be found in the build log, grepping for cmake. Use these options and follow the general build process	✅ success
Compatibility check	Checks that clickhouse binary runs on distributions with old libc versions. If it fails, ask a maintainer for help	✅ success
Docker keeper image	The check to build and optionally push the mentioned image to docker hub	✅ success
Docker server image	The check to build and optionally push the mentioned image to docker hub	✅ success
Docs check	Builds and tests the documentation	✅ success
Fast test	Normally this is the first check that is ran for a PR. It builds ClickHouse and runs most of stateless functional tests, omitting some. If it fails, further checks are not started until it is fixed. Look at the report to see which tests fail, then reproduce the failure locally as described here	✅ success
Flaky tests	Checks if new added or modified tests are flaky by running them repeatedly, in parallel, with more randomization. Functional tests are run 100 times with address sanitizer, and additional randomization of thread scheduling. Integrational tests are run up to 10 times. If at least once a new test has failed, or was too long, this check will be red. We don't allow flaky tests, read the doc	✅ success
Install packages	Checks that the built packages are installable in a clear environment	✅ success
Integration tests	The integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests	✅ success
Mergeable Check	Checks if all other necessary checks are successful	✅ success
PR Check	There's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS	✅ success
Stateful tests	Runs stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stress test	Runs stateless functional tests concurrently from several clients to detect concurrency-related errors	✅ success
Style check	Runs a set of checks to keep the code style clean. If some of tests failed, see the related log from the report	✅ success
Unit tests	Runs the unit tests for different release types	✅ success
Upgrade check	Runs stress tests on server version from last release and then tries to upgrade it to the version from the PR. It checks if the new server can successfully startup without any errors, crashes or sanitizer asserts	✅ success

[azat]

Test failures does not looks related:

Stateless tests (debug) [2/5] — fail: 1, passed: 1164, skipped: 10

• 01459_manual_write_to_replicas

Stateless tests (debug) [3/5] — fail: 2, passed: 1237, skipped: 22

• 01508_partition_pruning_long
• 02125_lz4_compression_bug

Upgrade check (debug) — Logical error thrown (see clickhouse-server.log or logical_errors.txt)
Upgrade check (tsan) — Logical error thrown (see clickhouse-server.log or logical_errors.txt)

• Logical error: not_processed should be empty #58319

[azat]

Stateless tests (tsan, s3 storage) [3/5] — fail: 1, passed: 1223, skipped: 38

• 02981_vertical_merges_memory_usage - 02981_vertical_merges_memory_usage flaky #59648 - tests: fix 02981_vertical_merges_memory_usage flakiness #59923

Timeout exceeded while receiving data from server. Waited for 300 seconds, timeout is 300 seconds

[devcrafter]

I think, the new behavior requires a setting. In general, writing into a currently read-only replication table is covered by insert retries.

The question is what semantic we want to bring with the new setting
(1) - prefer insert into currently non-read-only replicas
or
(2) - use only non-read-only replicas

AFAIU, the current implementation is about (1)

[devcrafter]

I think, the new behavior requires a setting. In general, writing into a currently read-only replication table is covered by insert retries.

The question is what semantic we want to bring with the new setting
(1) - prefer insert into currently non-read-only replicas
or
(2) - use only non-read-only replicas

AFAIU, the current implementation is about (1)

[azat]

I think, the new behavior requires a setting. In general, writing into a currently read-only replication table is covered by insert retries.

My thinking was that since it only prefers non-read-only replicas, there is no need in a separate setting, but I'm OK with adding one.

The question is what semantic we want to bring with the new setting
(1) - prefer insert into currently non-read-only replicas
or
(2) - use only non-read-only replicas

And yes, if there will be a setting then we can add a enum for it, i.e. prefer, no, yes or similar

AFAIU, the current implementation is about (1)

Yes.

[devcrafter]

AFAIU, the current implementation is about (1)

Yes.

Let's keep it. Doubt that we need (2) actually

[azat]

Let's keep it. Doubt that we need (2) actually

Do you still think that extra setting make sense?

[devcrafter]

Let's keep it. Doubt that we need (2) actually

Do you still think that extra setting make sense?

Hard to say, but at least there will be a way to control behavior. I guess there could be cases which benefit or not from this behavior (I guess it's mostly about load distribution). For example, if several long-running insert queries are thrown while some replicas will reconnect to zk, the inserts will be assigned to non-read-only replicas, i.e. load could not be distributed well.

When the user definitely knows that some replicas will be read-only for a long time, then it is worth excluding read-only replicas upfront. But then, we're talking about option (2) 🤔

[azat]

Okay, since you are also struggling about new setting for this functionality, I've added enum, distributed_insert_replicas_preferences=PREFER_NON_READ_ONLY/NO_PREFERENCES.

And also addressed all of your comments.

[azat]

Test failures does not looks related:

Stress test (debug) — Killed by signal (in clickhouse-server.log)
Stress test (tsan) — Killed by signal (in clickhouse-server.log)

• Logical error: Invalid storage definition in metadata file: it's a bug or result of manual intervention in metadata files #60349

[devcrafter]

It's a bit strange to introduce getManyCheckedForInsert (get many) which only used with PoolMode::GET_ONE (get one). But it looks fine since requires minimal changes in implementation.