Bryan Boreham

Gentle ping.

I believe the CI failure would be fixed by #14100, if you would like to rebase.

Looking at the code again, it seems to rely on a 10ms `Sleep` being reasonably close to 10ms. https://github.com/prometheus/prometheus/blob/3b8b57700c469c7cde84e1d8f9d383cb8fe11ab0/rules/manager_test.go#L2075 https://github.com/prometheus/prometheus/blob/3b8b57700c469c7cde84e1d8f9d383cb8fe11ab0/rules/manager_test.go#L1926 This assumption is not safe: https://github.com/golang/go/issues/44343 (but will be safer...

> 20M timeseries stored in the TSDB I've seen much bigger, without hitting this limit. So I suspect you hit it when Prometheus was trying to compact smaller blocks into...

#14200 seems to address some or all of the requirement; please let us know what remains.

Agreed, I saw someone confused over this yesterday. "user" is perhaps justifiable but confusing; "org" is better but confusing if one organisation runs multiple systems, e.g. dev and prod. The...

Some places you will see "user" vs "org" was copy-paste four years ago from code where those two things mean something; however in Cortex today they are never used to...

Was this fixed by #4341 ?

Answering my own question: no it was not fixed. That part improves, but `mergeIterator` seems to be so much slower that it defeats the improvement.

You’re right there isn’t a lot going on. I think I’m the only person listed in the MAINTAINERS file who’s still interested, though sadly I didn’t spend any time on...