Skip to content

monitoring: sync dot-com alerts to generator

Administrator requested to merge monitoring/more-out-of-band-alerts into master

Created by: bobheadxi

Syncs additional alerts from deploy-sourcegraph-dot-com, missed in https://github.com/sourcegraph/sourcegraph/issues/12117:

  • ProdLandingPageLatencyGoal -> removed - seems similar to page_load_latency? But updated page_load_latency to exclude GraphQL requests
  • ProdBlobLatencyGoal -> blob_load_latency, updated page_load_latency to exclude blob and give it a more realistic critical threshold (?) based on the threshold set for ProdBlobLatencyGoal
  • SearcherErrorRatioTooHigh -> removed because seems similar to unindexed_search_request_errors, also removed error_ratio because it seems very similar to unindexed_search_request_errors as well (?)
  • SearcherHighTraffic -> replica_traffic, inlined the record into the alert

See https://github.com/sourcegraph/deploy-sourcegraph-dot-com/pull/3099 for deploy-sourcegraph-dot-com change

Try it out

same as in https://github.com/sourcegraph/sourcegraph/pull/12391:

  1. kubectl port-forward svc/prometheus 9090:30090 -n prod
  2. Check out monitoring/migrate-out-of-band-alerts
  3. go generate ./monitoring
  4. ./dev/grafana.sh
  5. http://localhost:3370

Merge request reports