[9.1] feat: allow fractional MatchingDelay for sub-second matcher throttling - #8711
Merged
aldbr merged 2 commits intoAug 4, 2026
Merged
Conversation
chrisburr
force-pushed
the
feat/fractional-matching-delay
branch
2 times, most recently
from
July 16, 2026 08:41
f74cfc3 to
137c4fc
Compare
chrisburr
marked this pull request as ready for review
July 16, 2026 16:34
chrisburr
force-pushed
the
feat/fractional-matching-delay
branch
from
July 16, 2026 20:46
23237cb to
0ff974c
Compare
MatchingDelay values were cast with int(), so the minimum spacing between matched jobs of a given type was 1 second (>=1 job/s). Sites needing a higher steady start rate (e.g. to smooth a spiky batch of a few hundred jobs into a couple of hundred per two minutes) could not express the required sub-second spacing. Parse the MatchingDelay path as float via a cast parameter on __extractCSData, so fractional seconds (e.g. 0.5) are honoured; DictCache already builds the TTL with timedelta(seconds=...), which accepts floats. RunningLimit continues to parse as int.
The matching-delay counter was armed at the very end of Matcher.selectJob, ~7 DB round-trips after the negative condition was read. Under the Matcher's concurrent request handling, every thread in that window reads the type as un-delayed and matches it, so bursts of jobs slip past the delay (observed ~15-25% of matches landing <0.1s apart at a site with a 1s delay). Move the arming to immediately after the match is confirmed Waiting, and reuse the JobType already fetched for the status check so no extra query is added: updateDelayCounters now accepts pre-fetched attributes and only queries the DB for any configured attribute that is missing. This shrinks the read->arm window from ~7 round-trips to the match itself, cutting the leak, and removes a query from the match path.
aldbr
force-pushed
the
feat/fractional-matching-delay
branch
from
August 4, 2026 13:39
0ff974c to
688e1c8
Compare
aldbr
approved these changes
Aug 4, 2026
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
3 participants
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.
MatchingDelay values were cast with
int(), so the minimum spacing between matched jobs of a given type was 1 second per Matcher instance (>=1 job/s). Sites needing a higher steady start rate (e.g. to smooth a spiky batch of a few hundred jobs into a couple of hundred per two minutes) could not express the required sub-second spacing.Parse the
MatchingDelaypath as float via a cast parameter on__extractCSData, so fractional seconds (e.g. 0.5) are honoured;DictCachealready builds the TTL withtimedelta(seconds=...), which accepts floats.BEGINRELEASENOTES
*WorkloadManagement
NEW: allow fractional MatchingDelay for sub-second matcher throttling
ENDRELEASENOTES