Bryan Boreham

> There have been a few fixes in https://github.com/prometheus/prometheus/pull/10991 which also affect this branch; I should move them here, and particularly the extra tests. I have done this; unfortunately in...

I think it would be good to split this PR up, get some of the easier bits merged so I don't have to keep rebasing so much on merge conflicts....

> I think it would be good to split this PR up, get some of the easier bits merged so I don't have to keep rebasing so much on merge...

> Metrics from a test instance: Can you explain (or point out where I missed the explanation) what these pictures show? Assume I know nothing about your environment, your dashboards,...

Thanks, that helps a bit. I am puzzled why your charts don't show a repeating pattern every 2 hours, which is the standard head compaction interval. Also puzzled if your...

I would like to propose a change to your benchmark, to do compression just once outside of the benchmark loop. This way, the benchmark is more focused on decompression and...

Given you changed `BenchmarkGzip`, you should probably state the difference before/after in this PR. The changes to `discovery/xds/kuma_mads.pb.go` (annotated `DO NOT EDIT`) and `tsdb/fileutil/flock_js.go` do not seem relevant to compression,...

Given you changed `BenchmarkGzip`, you should probably state the difference before/after in this PR. The changes to `discovery/xds/kuma_mads.pb.go` (annotated `DO NOT EDIT`) and `tsdb/fileutil/flock_js.go` do not seem relevant to compression,...

Given you changed `BenchmarkGzip`, you should probably state the difference before/after in this PR. The changes to `discovery/xds/kuma_mads.pb.go` (annotated `DO NOT EDIT`) and `tsdb/fileutil/flock_js.go` do not seem relevant to compression,...

> > I didn't see any explanation in the description or commit message why they are necessary. > That was in a commit where I run gofumpt. Ok that's _what_...