Skip to content

HTTP connections pool#58845

Merged
CheSema merged 5 commits intomasterfrom
chesema-session-pool
Mar 10, 2024
Merged

HTTP connections pool#58845
CheSema merged 5 commits intomasterfrom
chesema-session-pool

Conversation

@CheSema
Copy link
Copy Markdown
Member

@CheSema CheSema commented Jan 16, 2024 •
edited
Loading

#50217

the main changes here are:

src/Common/HostResolvePool.{h.cpp}

That class resolves host into multiply addresses
Features:

  • balance address usage. selectBest() chooses the address by random with weights. The more ip is used the lesser weight it has. When new address is happened, it takes more weight. But still not all requests are assigned to the new address.
  • join resolve results. In case when host is resolved into different set of addresses, this class join all that addresses and use them. An address expires after history_ time.
  • failed address pessimization. If an address marked with setFail() it is marked as faulty. Such address won't be selected until either a) it still occurs in resolve set after history_ time or b) all other addresses are pessimized as well.
  • resolve schedule. Addresses are resolved through DB::DNSResolver::instance(). Usually it does not happen more often than once in history_ time. But also new resolve performed each setFail() call.

src/Common/ConnectionPool.{h,cpp}

That classes manage connections to the endpoint
Features:

  • it uses HostResolvePool for address selecting. See Common/HostResolvePool.h for more info.
  • it minimizes number of Session::connect()/Session::reconnect() calls
  • stores only connected and ready to use sessions
  • connection could be reused even when limits are reached
  • soft limit, warn limit, hard limit
  • Session::reconnect() uses the pool as well
  • comprehensive sensors
  • session is reused according its inner state, automatically
  • sensors and limits are groped by the logical source of connection: disks, storages, http (the rest).

src/IO/ReadWriteBufferFromHTTP.h

I have removed all unnecessary complications from it. Make more straightforward errors/retries handling.

Changelog category (leave one):

  • Improvement

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

This PR makes http/https connections reusable for all uses cases. Even when response is 3xx or 4xx.

@CheSema CheSema force-pushed the chesema-session-pool branch 2 times, most recently from a08e9d6 to 025688d Compare January 16, 2024 21:20
@robot-clickhouse robot-clickhouse added the pr-not-for-changelog This PR should not be mentioned in the changelog label Jan 16, 2024
@robot-clickhouse
Copy link
Copy Markdown
Member

robot-clickhouse commented Jan 16, 2024 •
edited by robot-ch-test-poll1
Loading

This is an automated comment for commit 77c5de7 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 nameDescriptionStatus
A SyncThere's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS✅ success
AST fuzzerRuns randomly generated queries to catch program errors. The build type is optionally given in parenthesis. If it fails, ask a maintainer for help✅ success
ClickBenchRuns [ClickBench](https://github.com/ClickHouse/ClickBench/) with instant-attach table✅ success
ClickHouse build checkBuilds 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 checkChecks that clickhouse binary runs on distributions with old libc versions. If it fails, ask a maintainer for help✅ success
Docker keeper imageThe check to build and optionally push the mentioned image to docker hub✅ success
Docker server imageThe check to build and optionally push the mentioned image to docker hub✅ success
Docs checkBuilds and tests the documentation✅ success
Fast testNormally 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 testsChecks 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 packagesChecks that the built packages are installable in a clear environment✅ success
Integration testsThe integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests✅ success
Mergeable CheckChecks if all other necessary checks are successful✅ success
PR CheckThere's no description for the check yet, please add it to tests/ci/ci_config.py:CHECK_DESCRIPTIONS✅ success
Performance ComparisonMeasure changes in query performance. The performance test report is described in detail here. In square brackets are the optional part/total tests✅ success
Stateful testsRuns stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc✅ success
Stress testRuns stateless functional tests concurrently from several clients to detect concurrency-related errors✅ success
Style checkRuns a set of checks to keep the code style clean. If some of tests failed, see the related log from the report✅ success
Unit testsRuns the unit tests for different release types✅ success
Upgrade checkRuns 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 nameDescriptionStatus
CI runningA 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
Stateless testsRuns stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc❌ failure

@CheSema CheSema force-pushed the chesema-session-pool branch 4 times, most recently from cb32f58 to 6930843 Compare January 16, 2024 23:35
@CheSema CheSema mentioned this pull request Jan 16, 2024
Closed
@CheSema CheSema force-pushed the chesema-session-pool branch 3 times, most recently from 5c5a05d to 91dcf5d Compare January 17, 2024 10:48
@CheSema CheSema mentioned this pull request Jan 19, 2024
Merged
@CheSema CheSema force-pushed the chesema-session-pool branch 12 times, most recently from ff9b4ef to 91bebb6 Compare January 25, 2024 21:39
@CheSema CheSema force-pushed the chesema-session-pool branch 5 times, most recently from f8d3d0d to af63d1c Compare February 4, 2024 23:53
@CheSema CheSema force-pushed the chesema-session-pool branch 4 times, most recently from 43d965a to 9b1b17b Compare March 2, 2024 19:53
@CheSema
Copy link
Copy Markdown
Member Author

CheSema commented Mar 3, 2024 •
edited
Loading

Found a bug:
A lot of files are read with default initialized read settings, files like format_version.txt, columns.txt, metadata_version.txt,... and indexes like primary.cidx. By default read settings has http_max_tries = 1, meanwhile if read settings created from default context it would be initializes with http_max_tries = 10.
As a quick fix I've made defaults corresponding.

It happens here:

ClickHouse/src/Storages/MergeTree/PartMetadataManagerOrdinary.cpp

Lines 11 to 14 in a1d1f76

std::unique_ptr<ReadBuffer> PartMetadataManagerOrdinary::read(const String & file_name) const
{
size_t file_size = part->getDataPartStorage().getFileSize(file_name);
auto res = part->getDataPartStorage().readFile(file_name, ReadSettings().adjustBufferSize(file_size), file_size, std::nullopt);

@CheSema CheSema force-pushed the chesema-session-pool branch 6 times, most recently from 376593a to cd52e32 Compare March 4, 2024 15:02
@CheSema CheSema requested a review from al13n321 March 4, 2024 16:12
@CheSema CheSema force-pushed the chesema-session-pool branch from cd52e32 to d330b0d Compare March 5, 2024 12:38
al13n321
al13n321 approved these changes Mar 7, 2024
View reviewed changes
al13n321
al13n321 reviewed Mar 7, 2024
View reviewed changes
@CheSema CheSema force-pushed the chesema-session-pool branch from d330b0d to 32b2c49 Compare March 7, 2024 13:49
@robot-ch-test-poll2 robot-ch-test-poll2 added the pr-improvement Pull request with some product improvements label Mar 7, 2024
@CheSema
Copy link
Copy Markdown
Member Author

CheSema commented Mar 8, 2024

Failed tests
#60625
#61090

@azat azat mentioned this pull request Mar 12, 2024
Merged
@Avogar Avogar mentioned this pull request Mar 12, 2024
Closed
@azat azat mentioned this pull request Mar 25, 2024
Merged
@CheSema CheSema mentioned this pull request Mar 28, 2024
Merged
30 tasks
@arthurpassos arthurpassos mentioned this pull request May 6, 2024
Merged
31 tasks
@Algunenano Algunenano mentioned this pull request May 29, 2024
Closed
@jmaicher jmaicher mentioned this pull request Nov 25, 2024
Open
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@al13n321 al13n321 al13n321 approved these changes

@alexey-milovidov alexey-milovidov alexey-milovidov left review comments

Assignees

@al13n321 al13n321

Labels

pr-improvement Pull request with some product improvements pr-synced-to-cloud The PR is synced to the cloud repo

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@CheSema @robot-clickhouse @al13n321 @alexey-milovidov @robot-ch-test-poll2 @robot-clickhouse-ci-1