Optimize query uniqueness check in ProcessList

[adikus]

Hi!
In our ClickHouse use case, we have a lot of users (currently ~couple thousands per CH instance), all of whom are expected to run queries.
Recently we tried to push this even further (~couple tens of thousands users per CH instance), which resulted in a significant degradation in our QPS throughput. We were able to achieve only between 100 and 200 QPS, with each query taking up to 30s (expected run time for these queries is under 100ms).

With tracing logs enabled, we were able to figure that most queries were spending their time on this lock:
https://github.com/ClickHouse/ClickHouse/blob/c2f4dc0f14f55cf9785da86cb2bf07c556e22d1e/src/Interpreters/ProcessList.cpp#L82C27-L82C27
The reason was that the inserts into the ProcessList were taking too long, specifically in this for loop:

ClickHouse/src/Interpreters/ProcessList.cpp

Lines 185 to 194 in c2f4dc0

	/// Check other users running query with our query_id
	for (const auto & user_process_list : user_to_queries)
	{
	if (user_process_list.first == client_info.current_user)
	continue;
	if (auto running_query = user_process_list.second.queries.find(client_info.current_query_id); running_query != user_process_list.second.queries.end())
	throw Exception(ErrorCodes::QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING,
	"Query with id = {} is already running by user {}",
	client_info.current_query_id, user_process_list.first);
	}

The for loop iterates over every user the server ever saw, which explains why most queries spend so much time on this lock (we saw times up to 10ms spend in the loop, which would explain the ~100 QPS limit we saw).
This PR, replaces the for loop with a single lookup into a std::unordered_map.

We tested this patch in our environment and achieved over 2000 QPS on the same workload as before (could definitely go higher).

Changelog category (leave one):

• Performance Improvement

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

Improve performance of executing queries for use cases with many users

Documentation entry for user-facing changes

• Documentation is written (mandatory for new features)

Information about CI checks: https://clickhouse.com/docs/en/development/continuous-integration/

[CLAassistant]

All committers have signed the CLA.

[robot-clickhouse]

This is an automated comment for commit f447c54 with description of existing statuses. It's updated for the latest CI running

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

Successful checks

Check name	Description	Status
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
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	✅ 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 image for servers	The check to build and optionally push the mentioned image to docker hub	✅ 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
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	✅ success
Push to Dockerhub	The check for building and pushing the CI related docker images to docker hub	✅ success
SQLTest	There's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS	✅ success
SQLancer	Fuzzing tests that detect logical bugs with SQLancer tool	✅ success
Sqllogic	Run clickhouse on the sqllogic test set against sqlite and checks that all statements are passed	✅ success
Stateful tests	Runs stateful 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

Check name	Description	Status
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	❌ failure

[ilejn]

I may be wrong, but at first glance we should erase from queries_to_user after cancelled_cv.wait(), otherwise we are going to have a race condition here.

[FrankChen021]

Hi @adikus , I'm wondering in your test, how many users are there? We also notice that the ClickHouse spends time on the ProcessList processing, in our case, there's only 1 user that issues queries up to 2K. I think the 1 user scenario cannot benefit much from this patch, right?

[jsc0218]

Hi @adikus , I'm wondering in your test, how many users are there? We also notice that the ClickHouse spends time on the ProcessList processing, in our case, there's only 1 user that issues queries up to 2K. I think the 1 user scenario cannot benefit much from this patch, right?

I don't think this patch benefits much in you case, since it mainly avoids the cost of scanning over the user_process_list, while in your case just one user.