Fix SIGSEGV due to CPU/Real profiler by azat · Pull Request #63865 · ClickHouse/ClickHouse

azat · 2024-05-15T16:30:50Z

The problem was due to incorrect unwinding due from signal handlers, which leads to incorrect DWARF (FDE/CIE) interpretation.

After this patch I was not able to reproduce the crash for couple of hours, while before it was very stable (I've reduced the minimal threshold for query_profiler_real_time_period_ns), using simply:

$ clickhouse-benchmark --port 19000 -q "SELECT * FROM remote('127.{1..10}', system, one)" --query_profiler_real_time_period_ns=1

Note, I'm using here remote() for fibers, that has stack with guard pages that helps with reproducing the crash more faster.

P.S. I also have another implementation of this fix, without patching unwind and using info from signal context directly, and even though it is better, because you don't need to trip extra frames and you can use all the 45 frames for something useful, it is too complex, so let's go with a simpler patch first, and I think it could be even backported.

Changelog category (leave one):

Bug Fix (user-visible misbehavior in an official stable release)

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

Fix SIGSEGV due to CPU/Real (query_profiler_real_time_period_ns/query_profiler_cpu_time_period_ns) profiler (has been an issue since 2022, that leads to periodic server crashes, especially if you were using distributed engine)

Fixes: #60460
Fixes: #33531
Fixes: #60219
Refs: ClickHouse/libunwind#25
Refs: llvm/llvm-project#92291

The problem was due to incorrect unwinding due from signal handlers, which leads to incorrect DWARF (FDE/CIE) interpretation. After this patch I was not able to reproduce the crash for couple of hours, while before it was very stable (I've reduced the minimal threshold for query_profiler_real_time_period_ns), using simply: $ clickhouse-benchmark --port 19000 -q "SELECT * FROM remote('127.{1..10}', system, one)" --query_profiler_real_time_period_ns=1 Note, I'm using here remote() for fibers, that has stack with guard pages that helps with reproducing the crash more faster. P.S. I also have another implementation of this fix, without patching unwind and using info from signal context directly, and even though it is better, because you don't need to trip extra frames and you can use all the 45 frames for something useful, it is too complex, so let's go with a simpler patch first, and I think it could be even backported. Signed-off-by: Azat Khuzhin <[email protected]>

robot-clickhouse-ci-2 · 2024-05-15T16:31:33Z

This is an automated comment for commit 94041d1 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
Integration tests	The integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests	❌ failure
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
Stateful tests	Runs stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	❌ failure
Stress test	Runs stateless functional tests concurrently from several clients to detect concurrency-related errors	❌ 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
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
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ 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

alexey-milovidov · 2024-05-16T03:16:34Z

@azat Let's fix all CI failures.

azat · 2024-05-16T07:49:08Z

Failures does not looks related, most of them are flaky

Integration tests (tsan) [1/6] — fail: 1, passed: 439

test_broken_projections/test.py::test_mutation_with_broken_projection - unrelated and flaky

Integration tests (tsan) [5/6] — fail: 1, passed: 574

test_storage_s3/test.py::test_seekable_formats_url - flaky

Stateful tests (release, ParallelReplicas) — fail: 1, passed: 131, skipped: 5

00088_global_in_one_shard_and_rows_before_limit - flaky

Stress test (debug) — Hung check failed, possible deadlock found (see hung_check.log)

Looks like some hang in - will take a look later

/build/src/IO/ParallelReadBuffer.cpp:290::DB::ParallelReadBuffer::finishAndWait()
/build/src/IO/ParallelReadBuffer.h:33::DB::ParallelReadBuffer::~ParallelReadBuffer()
/build/src/IO/ParallelReadBuffer.h:33::DB::ParallelReadBuffer::~ParallelReadBuffer()

Fixed in Fix deadlock in ParallelReadBuffer #63814

Performance Comparison [2/4] — 2 errors, 2 too long, 1 faster, 1 unstable

The error for the query is

clickhouse_driver.errors.ServerException: Code: 159.
select_format.query4.run0: DB::Exception: Timeout exceeded: elapsed 19.931821772 seconds, maximum: 15. Stack trace:

But it fails on the left server anyway (i.e. upstream/master), and sometimes it's execution took ~15 seconds, will take a look later.

The query is INSERT INTO table_PrettySpace SELECT * FROM test.hits LIMIT 10000, the problem is likely due to:

Better rendering of multiline strings in Pretty formats #59940 (comment)

filimonov · 2024-05-28T10:31:31Z

Backports?

azat · 2024-05-28T11:49:17Z

Agree, make sense (it should not make anything worse)

Backport #63865 to 24.3: Fix SIGSEGV due to CPU/Real profiler

Backport #63865 to 24.4: Fix SIGSEGV due to CPU/Real profiler

Backport #63865 to 24.2: Fix SIGSEGV due to CPU/Real profiler

…d update) * ClickHouse/ClickHouse#63865 from azat/fix-query-profiler-SIGSEGV Fix SIGSEGV due to CPU/Real profiler * ClickHouse/ClickHouse#60468 from ClickHouse/libunwind-fix-crash Fix crash in libunwind while interpreting debug info * ClickHouse/ClickHouse#65509 Update libunwind to 18.1.7 * ClickHouse/ClickHouse#66850 from ClickHouse/revert-libunwind-patch Revert libunwind patch * ClickHouse/ClickHouse#66977 from ClickHouse/uwo Apply libunwind fix * ClickHouse/ClickHouse#68312 from ClickHouse/muslwind Apply libunwind changes needed for musl * ClickHouse/ClickHouse#76107 from ClickHouse/owo Apply libunwind fix for DwarfFDECache * ClickHouse/ClickHouse#76136 from ClickHouse/revert-76107-owo Revert "Apply libunwind fix for DwarfFDECache" * ClickHouse/ClickHouse#76178 from ClickHouse/unw Apply libunwind fix for DwarfFDECache, attempt 2 * Update libunwind to aec8e58 * ClickHouse/ClickHouse#67152 from ClickHouse/ohno Uncomment accidentally commented out code in QueryProfiler * ClickHouse/ClickHouse#64058 aarch64 sigaltstack size fix * ClickHouse/ClickHouse#58607 Added null guards to avoid potential demangle crashes --------- Co-authored-by: Nikita Mikhaylov <[email protected]> Co-authored-by: Alexey Milovidov <[email protected]> Co-authored-by: Michael Kolupaev <[email protected]> Co-authored-by: Antonio Andelic <[email protected]> Co-authored-by: Max Kainov <[email protected]>

…d update) (#11324) * ClickHouse/ClickHouse#63865 from azat/fix-query-profiler-SIGSEGV Fix SIGSEGV due to CPU/Real profiler * ClickHouse/ClickHouse#60468 from ClickHouse/libunwind-fix-crash Fix crash in libunwind while interpreting debug info * ClickHouse/ClickHouse#65509 Update libunwind to 18.1.7 * ClickHouse/ClickHouse#66850 from ClickHouse/revert-libunwind-patch Revert libunwind patch * ClickHouse/ClickHouse#66977 from ClickHouse/uwo Apply libunwind fix * ClickHouse/ClickHouse#68312 from ClickHouse/muslwind Apply libunwind changes needed for musl * ClickHouse/ClickHouse#76107 from ClickHouse/owo Apply libunwind fix for DwarfFDECache * ClickHouse/ClickHouse#76136 from ClickHouse/revert-76107-owo Revert "Apply libunwind fix for DwarfFDECache" * ClickHouse/ClickHouse#76178 from ClickHouse/unw Apply libunwind fix for DwarfFDECache, attempt 2 * Update libunwind to aec8e58 * ClickHouse/ClickHouse#67152 from ClickHouse/ohno Uncomment accidentally commented out code in QueryProfiler * ClickHouse/ClickHouse#64058 aarch64 sigaltstack size fix * ClickHouse/ClickHouse#58607 Added null guards to avoid potential demangle crashes --------- Co-authored-by: Nikita Mikhaylov <[email protected]> Co-authored-by: Alexey Milovidov <[email protected]> Co-authored-by: Michael Kolupaev <[email protected]> Co-authored-by: Antonio Andelic <[email protected]> Co-authored-by: Max Kainov <[email protected]>

robot-clickhouse-ci-2 added pr-bugfix Pull request with bugfix, not backported by default submodule changed At least one submodule changed in this PR. labels May 15, 2024

This was referenced May 15, 2024

libunwind: getSavedRegister: invalid read from memory #60460

Closed

Fix crash in libunwind while interpreting debug info #60468

Merged

nikitamikhaylov self-assigned this May 16, 2024

nikitamikhaylov added this pull request to the merge queue May 16, 2024

nikitamikhaylov approved these changes May 16, 2024

View reviewed changes

Merged via the queue into ClickHouse:master with commit ee984e8 May 16, 2024

azat deleted the fix-query-profiler-SIGSEGV branch May 16, 2024 13:14

robot-clickhouse-ci-1 added the pr-synced-to-cloud The PR is synced to the cloud repo label May 16, 2024

alesapin mentioned this pull request May 16, 2024

Not a bug: libunwind: aarch64: SIGSEGV signal handler leads to another seg fault #63855

Closed

alexey-milovidov added the pr-must-backport Pull request should be backported intentionally. Use this label with great care! label May 29, 2024

This was referenced May 29, 2024

Cherry pick #63865 to 23.8: Fix SIGSEGV due to CPU/Real profiler #64555

Closed

Cherry pick #63865 to 24.2: Fix SIGSEGV due to CPU/Real profiler #64556

Merged

robot-clickhouse added a commit that referenced this pull request May 29, 2024

Backport #63865 to 24.2: Fix SIGSEGV due to CPU/Real profiler

597ad76

This was referenced May 29, 2024

Backport #63865 to 24.2: Fix SIGSEGV due to CPU/Real profiler #64557

Merged

Cherry pick #63865 to 24.3: Fix SIGSEGV due to CPU/Real profiler #64558

Merged

robot-clickhouse added a commit that referenced this pull request May 29, 2024

Backport #63865 to 24.3: Fix SIGSEGV due to CPU/Real profiler

084a7f4

This was referenced May 29, 2024

Backport #63865 to 24.3: Fix SIGSEGV due to CPU/Real profiler #64559

Merged

Cherry pick #63865 to 24.4: Fix SIGSEGV due to CPU/Real profiler #64560

Merged

robot-clickhouse added a commit that referenced this pull request May 29, 2024

Backport #63865 to 24.4: Fix SIGSEGV due to CPU/Real profiler

6fbcccb

robot-ch-test-poll3 mentioned this pull request May 29, 2024

Backport #63865 to 24.4: Fix SIGSEGV due to CPU/Real profiler #64561

Merged

robot-ch-test-poll3 added the pr-backports-created-cloud deprecated label, NOOP label May 29, 2024

robot-ch-test-poll2 added a commit that referenced this pull request May 29, 2024

Merge pull request #64559 from ClickHouse/backport/24.3/63865

bee2a57

Backport #63865 to 24.3: Fix SIGSEGV due to CPU/Real profiler

nikitamikhaylov added a commit that referenced this pull request May 29, 2024

Merge pull request #64561 from ClickHouse/backport/24.4/63865

ffed48c

Backport #63865 to 24.4: Fix SIGSEGV due to CPU/Real profiler

nikitamikhaylov added a commit that referenced this pull request May 29, 2024

Merge pull request #64557 from ClickHouse/backport/24.2/63865

bc6d77d

Backport #63865 to 24.2: Fix SIGSEGV due to CPU/Real profiler

robot-ch-test-poll added the pr-backports-created Backport PRs are successfully created, it won't be processed by CI script anymore label May 29, 2024

azat mentioned this pull request Jun 6, 2024

Not a bug: ARM64 crash when unwinding the stack #64915

Closed

shiyer7474 added a commit to Altinity/ClickHouse that referenced this pull request Sep 2, 2024

23.8 Backport of ClickHouse#63865 Fix SIGSEGV in libunwind

d680ee2

This was referenced Sep 2, 2024

23.8 Backport of #63865 Fix SIGSEGV in libunwind Altinity/ClickHouse#452

Closed

23.8.16 Backport PR #63685 Fix SIGSEGV due to CPU/Real profiler Altinity/ClickHouse#453

Closed

shiyer7474 mentioned this pull request Sep 15, 2024

23.8 Backport #63865 fix SIGSEGV due to profiler to 23.8.16 Altinity/ClickHouse#466

Merged

Enmk mentioned this pull request Sep 26, 2024

23.8.16 Pre-release Altinity/ClickHouse#436

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix SIGSEGV due to CPU/Real profiler#63865

Fix SIGSEGV due to CPU/Real profiler#63865
nikitamikhaylov merged 1 commit intoClickHouse:masterfrom
azat:fix-query-profiler-SIGSEGV

azat commented May 15, 2024 •

edited

Loading

Uh oh!

robot-clickhouse-ci-2 commented May 15, 2024 •

edited by robot-ch-test-poll3

Loading

Uh oh!

alexey-milovidov commented May 16, 2024

Uh oh!

azat commented May 16, 2024 •

edited

Loading

Uh oh!

filimonov commented May 28, 2024

Uh oh!

azat commented May 28, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

Conversation

azat commented May 15, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changelog category (leave one):

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

Uh oh!

robot-clickhouse-ci-2 commented May 15, 2024 • edited by robot-ch-test-poll3 Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

alexey-milovidov commented May 16, 2024

Uh oh!

azat commented May 16, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

filimonov commented May 28, 2024

Uh oh!

azat commented May 28, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

azat commented May 15, 2024 •

edited

Loading

robot-clickhouse-ci-2 commented May 15, 2024 •

edited by robot-ch-test-poll3

Loading

azat commented May 16, 2024 •

edited

Loading