| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
There was a problem hiding this comment.
Thanks for the follow-up — the ownership accounting for the three failure paths checks out (I traced the payload/ctx/permit/memory balances per path, and ran ProducerMemoryLeakTest on this branch locally: 22/22 pass).
The inline comments below are structural/consistency suggestions rather than correctness issues — no blockers from my side. One non-code nit: the PR description still contains template residue (the "trivial rework / covered by existing tests / added tests" options and the example verification bullets) — worth cleaning up before merge.
Review assisted by Claude Code (AI-assisted review), posted by @nodece.
Sorry, something went wrong.
|
@nodece Just made a fix, please review it again. |
Sorry, something went wrong.
|
The flaky issue has been fixed. |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Motivation
This follows up on #26455 to address three remaining producer send-failure cases. An oversized message serialized after schema registration can be cleaned up twice if its failure callback closes the producer during reconnect. A rejected chunk write can leave the send loop building subsequent chunks after the message has already failed, causing repeated callbacks and buffer releases. Interrupting a blocking chunked send also leaves references held for chunks that were never created.
Modifications
Verifying this change
(Please pick either of the following options)
This change is a trivial rework / code cleanup without any test coverage.
(or)
This change is already covered by existing tests, such as (please describe tests).
(or)
This change added tests and can be verified as follows:
(example:)
Does this pull request potentially affect one of the following parts:
If the box was checked, please highlight the changes