Philip Stoev comments

Results 64 comments of


                                            Philip Stoev

[Epic] Improve interactive query times

Related tickets from times past: - #6738 - #3740

[Epic] Improve interactive query times

Tickets about loss of interactivity due to issues in the dataflow layer. I guess those also count towards improved interactive query times: - #8853 - #7577 - #7575

performance: spikes in latency (>0.5-1s) for simple , non-concurrent SELECT queries

How do they fail exactly?

performance: spikes in latency (>0.5-1s) for simple , non-concurrent SELECT queries

so the query returns no rows with `serializable` ? That is not good

performance: spikes in latency (>0.5-1s) for simple , non-concurrent SELECT queries

oh so tables are now hm "eventually consistent" under `serializable` TIL

regression: storaged crash loop: some element of the Sink as_of frontier is too far advanced for our output-gating timestamp"

Happened again https://buildkite.com/materialize/nightlies/builds/1175#_

pg-cdc: Duplicates in source if storaged is killed and restarted

I am adding the C-Triage tag to this bug in order for it to enter back into the triage system. There has been zero movement (tags, priority, assigment, anything) on...

pg-cdc: Duplicates in source if storaged is killed and restarted

I rebased and the test no longer fails, so I merged it to "main". I am leaving this ticket open in case any further discussion and/or fixes needs to happen.

WIP: suss out problems arising from storaged parallelism

Yes, I have a fix for the `mz_materializations` , so please disregard those for the time being.

WIP: suss out problems arising from storaged parallelism

When ready, please push only our fix and none of the changes needed to get the --workers 4 test running. I have a separate branch that achieves that but in...