-
Notifications
You must be signed in to change notification settings - Fork 591
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: table scan with too many delete tombstones is slow #12903
Comments
Is this generalizable to some sort of cache? Such that we can handle tombstone ranges.
I think @Li0k has worked on this. |
Can I confirm again @Li0k? This is done by your PRs right? |
yeah, but cannot solve it completely, If the tombstone is affecting performance, we can consider adjusting the frequency of the tombstone picker triggers |
Further discussed offline with @Li0k , this part is still under discussion.
This part is done, compaction will occur in 10 min intervals. See #12776. |
For testing compaction: https://github.com/risingwavelabs/kube-bench/issues/306 |
This issue has been open for 60 days with no activity. Could you please update the status? Feel free to continue discussion or close as not planned. |
To resolve the backfill live-lock described in #12680, we ensure that the backfill should make progress before getting canceled in #12780. However, the underlying performance is still not resolved. This may affect the scan performance on tables with frequent deletes, like the materialized views cleaned up with the temporal filter.
Originally posted by @BugenZhao in #12680 (comment)
This can be easily reproduced:
As expected, nearly all time is spent on rewinding the merge iterator.
Ideas given by @hzxa21:
The text was updated successfully, but these errors were encountered: