Łukasz Mierzwa

Isn’t the same design breaking behaviour already supported in Prometheus by metrics relabel rules? One can drop arbitrary time series there, resulting in the same issue you describe.

Rebased on top of current main branch

Any chance this be moved forward now?

> I was reviewing the PR, but I stopped after noticing something: are we m-mapping only right before the compaction or am I missing someplace where the m-mapping is triggered...

Initially fieldalignment results looked as if there was a lot of memory being wasted due to padding. But that's only because it reports two things: - memory wasted due to...

> You don't mention it explicitly, so I will note that sorting is disabled during WAL replay. Didn't notice that, thanks.

This is a bit tricky to benchmark because it's a mixture of concurrency and also sorting (differently ordered data might cost different amount of cpu time to sort). But I...

This is unlikely to be a bug in Prometheus but most likely problem on your end. If you look at timestamps you’ll notice that they are duplicated. There are always...

Handle what exactly? In Prometheus you’re supposed to have unique labels on all time series. Automatic injection of job and instance labels usually ensures this. So first you need to...

If you use thanos and that’s where you see this problem then maybe thanos is merging two counters from two different Prometheus servers into a single time series? Try your...