Process metrics dashboard #1109
Conversation
mariomac
requested review from
grcevski,
marctc and
rafaelroquetto
as code owners
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1109 +/- ##
==========================================
- Coverage 81.96% 79.20% -2.76%
==========================================
Files 141 136 -5
Lines 11595 11293 -302
==========================================
- Hits 9504 8945 -559
- Misses 1563 1803 +240
- Partials 528 545 +17
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
grafana/process-dashboard.json
Outdated
| { | ||
| "datasource": { | ||
| "type": "prometheus", | ||
| "uid": "${DS_GRAFANACLOUD-MARIOMACIAS-PROM}" |
There was a problem hiding this comment.
Maybe we want to rename this something more generic like ${datasource}
| @@ -0,0 +1,556 @@ | |||
| { | |||
| "__inputs": [ | |||
There was a problem hiding this comment.
I didn't see this section in our other dashboards, so maybe is not needed?
There was a problem hiding this comment.
Yeah. This dashboard is exported with the "Export for sharing externally" option, while the others don't. I'm not sure which is the correct form. I tried exporting/importing in both formats, and the difference is that, during the import, this form allows choosing the source in the import dialog.
* Divided disk and network panels in send/receive * Reformat legend
* First snapshot of process metrics dashboard * make it work with prometheus-scraped metrics * * Addressed Marc's comments * Divided disk and network panels in send/receive * Reformat legend * revert unwanted changes in dashboard.json
It works with both OTEL-generated metrics (which sends resource attributes separated in the target_info metric) and Prometheus-scraped metrics (which send resource attributes in the same metric). The screenshot shows the graphs split in two parts: the left part is capturing the Prometheus metrics, and the right part is after switching Beyla to export via OTEL endpoint.
It achieves that by having two types of queries:
process_pid=""(as the OTEL exporter does not send this attribute in the metric), joined withtarget_info{process_pid!=""}metricprocess_pid!=""(as the Prometheus exporter includes this attribute in the metric)So, to let this dashboard work, the user should let the
process_pidattribute enabled (default).