-
Story
-
Resolution: Done
-
Normal
-
None
-
None
-
None
The queries are there under Alerts in grafana.
Stephen recently did similar for component readiness where sippy periodically queries bigquery and publishes metrics. We want to do similar with these disruption queries. Refresh every 12 hours. (for both actually) Use cache that he established as well for a second layer of protection against repeat queries.
Push out the metrics with appropriate labels.
Add alerts in DPCR to alert if values are above expected for three days.
- blocks
-
TRT-1105 Design system for preventing disruption regressions in the distribution tail
- Closed
- links to