Closed owent closed 4 months ago
@ThomsonTan Is there any problem in this PR? Maybe we can close #2553 if this PR is merged. It fixes the same problems as #2553 do and also fix the simular problems in trace and logs.
BTW, can you explain in brief that why the sequence notification can break the potential forever wait? Thanks.
BTW, can you explain in brief that why the sequence notification can break the potential forever wait? Thanks.
When more than one threads call ForceFlush
concurrency, is_force_flush_notified
may be set true
only once in background thread and one of the threads call ForceFlush
will wait is_force_flush_notified
for ever.
Attention: Patch coverage is 95.94595%
with 3 lines
in your changes are missing coverage. Please review.
Project coverage is 87.70%. Comparing base (
497eaf4
) to head (d3df9a4
). Report is 65 commits behind head on main.
Adding ok-to-merge.
Waiting to give @lalitb and @esigo a chance to comment, and planning to merge early next week.
Fixes #2574
May also fixes #2583
Changes
ForceFlush
.For significant contributions please make sure you have completed the following items:
CHANGELOG.md
updated for non-trivial changes