Marco Pracucci

Next steps: - Wait to rollout new `Series()` latency breakdown metric to production and see how much the `Send()` impacts there - Serialize messages before calling `Send()` and then send...

> Wait to rollout new Series() latency breakdown metric to production and see how much the Send() impacts there We're rolling out the new metrics to production and we have...

> Action item: ss suggested by @fpetkovski I'm now even more convinced we should measure the actual time taken by encoding the message and the Send() separately, so we have...

> I checked the value by PromQL, the result is 42.3%. Could I see the result of the query run as a range query around the time the alert fired?

This looks a nice idea! I'm fine keeping the link enabled for anyone. There's nothing bad being opinionated about the best logs database in the world 😉

@osg-grafana What's up with this PR? Is there any plan to move forward? Otherwise we may close it for now.

The ruler API endpoints don't support tenant federation. Tenant federation is currently supported only be query API endpoints. Please try to specify only 1 single tenant in the datasource used...

I see two different issues in the example logs. > sample at timestamp 1673248520000 (2023-01-09 07:15:20 +0000 UTC) has value -2076.414045 while was expecting -2079.116897 Both logs of this type...

I currently don't have the bandwidth to follow up on this. To further debug it, I would try to add more details to the log when the query fail to...

> But limiting chunks is less trivial. We can maybe estimate the number of chunks each series will have based on the numbers of chunks per series in the first...