/deploy-to-hg

• Preparing your instance. A comment containing your instance's url will be added to this PR when the instance is ready.
• Your instance will be ready in ~10 minutes. Follow the workflow progress
• Slack channel: #proj-ephemeral-hg-instances
• Building instance with andre/traceql-metrics-exemplars oss branch and main enterprise branch. How to choose a branch

Error building instance: Contact #proj-ephemeral-hg-instances if it is not a compile error. Logs

This is really cool, nice job 👍

Tooltips seems to get stuck in the UI after hovering over a few exemplars.

If I click around a few times on the various tabs + RED metric cards, eventually the diamonds turn into x's.

Do you know for certain why some exemplars are darker than others?

The upper/lower case combinations between key/value pairs in the tooltip are a little confusing. I can see the reasoning but this will not be as clear to users. We can revisit this with some UX help in the New Year.

We don't have much control over the rendering portion here since the exemplars are being rendered by a plugin of the Timeseries and Heatmap visualizations. So I can't really explain why some of these things happen, but if they do we can either try to fix them in another PR or create issues for the owner of the vizualisations to take a look. However, these exemplars are the same as the Prometheus exemplars.

Upon changing tabs, all of the exemplars disappear from the rate metric card.

I'll take a look!

How is the value from the exemplar reflected in the trace? Is it just the case that this trace was recorded during the course of this rate (error) of requests? (this is to solidify my own understanding)

Yes, Tempo will take a sample of the traces that generated these metrics and link them to the series. Tempo sends the trace ID and a few bits of metadata (like the timestamp and value) so the exemplar can be rendered in the right point of the graph.