Marco Pracucci
Marco Pracucci
I've spent the last working day benchmarking "ring on Consul" performances. To do it, I've built a tiny tool to run X lifecycles and Y clients ([code](https://github.com/grafana/dskit/compare/load-test-consul)) and run it...
The `KVConfig` defined in `kv/memberlist/memberlist_client.go` includes `MetricsRegisterer prometheus.Registerer`: https://github.com/grafana/dskit/blob/a48e5b8c20f9abb78b349821b04a78cf2a40ad73/kv/memberlist/memberlist_client.go#L157-L159 This looks an anti-pattern to me. The registerer is typically specified in the NewXXX() function. In this case, I'm wondering if...
I just realised that the `kgo.RequestTimeoutOverhead()` is the default timeout for most of the requests. I didn't understand it just reading the doc, as I thought it was always an...
In this PR I'm proposing to add `req` field to 'read from broker errored' log. While investigating an incident, we had a case where such addition would have been helpful....
In [Mimir](https://github.com/grafana/mimir), we have a process creating 1 `Client` (setting `MetadataMinAge` = MetadataMaxAge = 10s`) and then using it to do two things: 1. Consume a partition 2. Request `ListOffsets`...
I propose to enrich the `read from broker errored, killing connection` log message adding the Kafka message type as an extra log field. Are you open to such change? If...
#### What this PR does _To be written..._ #### Which issue(s) this PR fixes or relates to N/A #### Checklist - [ ] Tests updated. - [ ] Documentation added....
Seen a flaky run of `TestSchedulerProcessor_processQueriesOnSingleStream` ([CI](https://github.com/grafana/mimir/actions/runs/8659078973/job/23744238911?pr=7882)): ``` --- FAIL: TestSchedulerProcessor_processQueriesOnSingleStream (1.65s) --- FAIL: TestSchedulerProcessor_processQueriesOnSingleStream/should_not_cancel_query_execution_if_scheduler_client_returns_a_non-cancellation_error (0.51s) scheduler_processor_test.go:253: Error Trace: /__w/mimir/mimir/pkg/querier/worker/scheduler_processor_test.go:253 Error: Not equal: expected: 2 actual : 1 Test: TestSchedulerProcessor_processQueriesOnSingleStream/should_not_cancel_query_execution_if_scheduler_client_returns_a_non-cancellation_error...
Last Satuday we had an incident caused by prometheus/prometheus#14525. The problem is that after #14525, the struct field `shardHashOrMemoryMappedMaxTime` is used for two different unrelated purposes. The value is updated...
## Proposal In November 2021, Grafana Labs [announced query sharding](https://grafana.com/blog/2021/11/09/observabilitycon-announcements/) for Grafana Enterprise Metrics (GEM). The query sharding implementation leverages some changes we did in TSDB to be able to...