Repository navigation
Fix flaky FPM test for bug #77023 - #24999
Merged
Merged
Conversation
The test sent two parallel requests to a pool with a single child and expected the dynamic process manager to spawn a second child. That only happens when the idle server maintenance heartbeat (1s) fires while the child is busy, but the scripts kept it busy for just 300ms each, so the spawn was a matter of luck and the total processes check failed on slow machines. Use a single request that blocks the child for longer than the heartbeat so the spare child is always spawned, and rely on the fact that the blocked child is the oldest idle one and is therefore quit first. One blocked child is enough to verify the SIGKILL fallback. Replace the fixed sleep with log timeouts and poll the status until the spare child is quit.
bukka
force-pushed
the
fpm-bug77023-test-flakiness
branch
from
October 28, 2026 13:10
35ddc2a to
5f1bcca
Compare
Member
Author
|
I double checked that it still fails with the original fix reverted and it does so the test still testing the issue. |
Member
Author
|
Just to note that this was causing most failures in our security repo so it should make the CI green there more often hopefully. |
bukka
added a commit
that referenced
this pull request
Oct 29, 2026
* PHP-8.2: Fix flaky FPM test for bug #77023 (#24999)
bukka
added a commit
that referenced
this pull request
Oct 29, 2026
* PHP-8.3: Fix flaky FPM test for bug #77023 (#24999)
bukka
added a commit
that referenced
this pull request
Oct 29, 2026
* PHP-8.4: Fix flaky FPM test for bug #77023 (#24999)
bukka
added a commit
that referenced
this pull request
Oct 29, 2026
* PHP-8.5: Fix flaky FPM test for bug #77023 (#24999)
bukka
added a commit
that referenced
this pull request
Oct 29, 2026
* PHP-8.6: Fix flaky FPM test for bug #77023 (#24999)
arnaud-lb
added a commit
to lazerg/php-src
that referenced
this pull request
Oct 29, 2026
* up/PHP-8.6: Fix json_encode() reporting false "Recursion detected" after OOM for hooked-property objects (php#24992) Fix phpGH-25023: ZEND_CONTAINER_OF() fails to compile with -std=c11 (php#25027) Fix flaky FPM test for bug #77023 (php#24999) Update GitHub Action workflows to `actions/checkout@v7` (8.6) Update GitHub Action workflows to `actions/checkout@v7` (8.5) Update GitHub Action workflows to `actions/checkout@v7` (8.4) Update GitHub Action workflows to `actions/checkout@v7` Fix phpGH-23626: OPcache: orphaned temporaries cause leaks and assertions [ci skip] Fix comment Fix block pass dropping ZEND_FAST_CALL OP2 definition Fix phpGH-23572: JIT calls wrong static method from trait clones Fix phpGH-24784: Preserve certificates when reading PKCS7/CMS CRLs (php#24785)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The test sent two parallel requests to a pool with a single child and expected the dynamic process manager to spawn a second child. That only happens when the idle server maintenance heartbeat (1s) fires while the child is busy, but the scripts kept it busy for just 300ms each, so the spawn was a matter of luck and the total processes check failed on slow machines.
Use a single request that blocks the child for longer than the heartbeat so the spare child is always spawned, and rely on the fact that the blocked child is the oldest idle one and is therefore quit first. One blocked child is enough to verify the SIGKILL fallback. Replace the fixed sleep with log timeouts and poll the status until the spare child is quit.