-
Notifications
You must be signed in to change notification settings - Fork 231
perf(coord): Timeout query immediately if it stayed in mailbox for too long #761
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
perf(coord): Timeout query immediately if it stayed in mailbox for too long #761
Conversation
query/src/main/scala/filodb/query/exec/MultiSchemaPartitionsExec.scala
Outdated
Show resolved
Hide resolved
d51d311
to
1db479e
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looking forward to this change. Would you mind adding comments about where the scoring is disabled? Not obvious in the code.... just comment so we are aware why the code is the way it is and the impact scoring has.
I am undoing most of this PR, retaining just the timeout checks. The JMH is not showing any improvement. It appears that Lucene's "score" methods actually do the job of search instead of scoring. Scoring is disabled already. I was able to debug through the steps and confirm it. |
2b813e5
to
4908e00
Compare
Pull Request checklist
Using certain
Occurs
clauses enables scoring/ranking of search results which appears to be CPU intensive in profiler results. This PR uses Occurs.FILTER that explicitly disables scoring and weighting of results.Also do timeout check before doing anything for query execution. This accounts for delays in delivery from actor mailbox.
TODO: