MDEV-33070 Thread pool starvation at oversubscribe#4804
MDEV-33070 Thread pool starvation at oversubscribe#4804gkodinov merged 1 commit intoMariaDB:10.11from
Conversation
gkodinov
left a comment
There was a problem hiding this comment.
Thank you for your contribution! This is a preliminary review.
vaintroub
left a comment
There was a problem hiding this comment.
Commented on test case. The one-liner code fix is ok, but test needs some improvement (and explanation)
3b1f419 to
dfd6a1d
Compare
vaintroub
left a comment
There was a problem hiding this comment.
Thank you! Looks quite good to me, nice explanation for the test case.
The only thing that I'd remove is a synchronous SELECT SLEEP(1), as commented. Perhaps you can slightly increase wait_timeout up from 2 seconds, when checking that INSERT went through(as commented, CI sometimes have weird timing issues), though I do not have a very strong opinion.
dfd6a1d to
7af1a72
Compare
vaintroub
left a comment
There was a problem hiding this comment.
Thanks for addressing my comments! Looks good to me now.
gkodinov
left a comment
There was a problem hiding this comment.
LGTM. Thanks for working on this.
gkodinov
left a comment
There was a problem hiding this comment.
Please rebase to 10.11: this is a bug fix and it needs to be in the lowest affected version.
|
I will rebase. and merge. |
7af1a72 to
aea0a9f
Compare
Allow one more worker at the oversubscribe threshold so a group only becomes oversubscribed after it exceeds the configured limit. Add a regression test that reproduces the starvation case in the generic thread pool and verifies queued work still drains. Signed-off-by: Varun Deep Saini <varun.23bcs10048@ms.sst.scaler.com>
aea0a9f to
3735d28
Compare
|
@gkodinov Thanks a lot !! ❤️ |
Summary
Testing