refactor: Reduce the time range of release info dashboard to 24h
We now refresh our release-related metrics with a scheduled pipeline, which means we don't need to fetch 22 days worth of data to display this dashboard. In grafana explore, we can tell that the metrics delivery_release_patch_status
and delivery_release_monthly_status
doesn't disappear with the pods restarting.
I refactored the file to filter for the latest-created metric, and the time range to be 24 hours (1 day). Hopefully this addresses gitlab-com/gl-infra/delivery#20181 (closed) where we were seeing latency loading the dashboard, and the delay between the emission and display of the metrics.
Snapshot dashboard (24 hours): https://dashboards.gitlab.net/dashboard/snapshot/5m6acmEAXaN19HIUeoCnXLU1gECoTr42?var-environment=gprd&orgId=1&var-PROMETHEUS_DS=mimir-gitlab-ops&var-stage=main
To generate a new snapshot (you need to replace PROMETHEUS_DS=
to mimir-gitlab-ops
, since the data lives in ops
mimir data source):
$ ./test-dashboard.sh delivery/release_info.dashboard.jsonnet
Installed https://dashboards.gitlab.net/dashboard/snapshot/5m6acmEAXaN19HIUeoCnXLU1gECoTr42?var-environment=ops&orgId=1&var-PROMETHEUS_DS=mimir-gitlab-ops&var-stage=main - delivery: Release Information
addressing: gitlab-com/gl-infra/delivery#20181 (closed)