Marco Pracucci
Marco Pracucci
> I pushed a change to use a specific flag instead of a `flavour`. > > There are now also some basic docs. I think we have to add significantly...
> Also updates mixtool to latest version which closes https://github.com/grafana/mimir/issues/1617 > Mixtool was actually updated two months ago, so the dashboards are already all compliant! Thanks for pointing to that...
Another occurrence [here](https://github.com/grafana/mimir/actions/runs/10555914541/job/29240417263?pr=9098): ``` --- FAIL: TestDispatcherGroupLimits (3.22s) --- FAIL: TestDispatcherGroupLimits/low_limit (3.03s) alertmanager_test.go:141: expected , got Diff: --- metric output does not match expectation; want +++ got: @@ -2,3 +2,3...
[another failure](https://github.com/grafana/mimir/actions/runs/9659364262/job/26642578055?pr=8511) ``` --- FAIL: TestMultitenantCompactor_DeleteLocalSyncFiles (6.16s) compactor_test.go:2003: Error Trace: /__w/mimir/mimir/pkg/compactor/compactor_test.go:2003 Error: Should not be zero, but was 0 Test: TestMultitenantCompactor_DeleteLocalSyncFiles ```
> there are the fields which show the number of bytes retried before storing in the cache - results_cache_hit_bytes and results_cache_miss_bytes. Doesn't this work for you? It doesn't tell if...
> was this solved by https://github.com/grafana/mimir/pull/4950? No. #4950 introduced a cache for the LIST objects API call, but it's (intentionally) just used when synching rule groups and not by ruler...
> Without this change, grpcurl complains that it cannot resolve mimir.proto. @pracucci, you added this tool in https://github.com/grafana/mimir/pull/6523 - does the script work for you without this change? I'm not...
I'm looking at the last reported flaky execution. We added a specific log to investigate this flaky test: ``` distributor_ingest_storage_test.go:463: Kafka topic test end offsets: kadm.ListedOffsets{"test":map[int32]kadm.ListedOffset{0:kadm.ListedOffset{Topic:"test", Partition:0, Timestamp:-1, Offset:0, LeaderEpoch:0,...
I suspect this test flakyness has the same root cause of https://github.com/grafana/mimir/issues/10693. Both tests, sometimes, apparently don't get all committed records when using the function `readAllMetricNamesByPartitionFromKafka()`. This function is used...