Ship additional GitLab Prometheus alerts (#45740) · Issues · GitLab.org / GitLab FOSS · GitLab (original) (raw)

Skip to content

GitLab Next

Ship additional GitLab Prometheus alerts

We will be shipping AlertManager via GitLab 10.8 (omnibus-gitlab#2999 (closed)). I think we should begin shipping default alerts for GitLab administrators. What metrics are most useful to add ASAP as alerts for most GitLab users/customers?

Some ideas:

Component Exporter Endpoint Prometheus metric
Unicorn http://localhost:8080/-/metrics unicorn_active_connections
Unicorn http://localhost:8080/-/metrics unicorn_queued_connections
Unicorn http://localhost:8080/-/metrics job_register_attempts_failed_total
Sidekiq http://localhost:9168/sidekiq sidekiq_queue_size
Gitaly http://localhost:9236/metrics grpc_server_handled_total{grpc_code="ResourceExhausted}
PostgreSQL http://localhost:9187/metrics pg_stat_database_deadlocks
PostgreSQL http://localhost:9187/metrics pg_stat_database_conflicts_confl_deadlock
Redis http://localhost:9121/metrics redis_up
Workhorse http://localhost:9229/metrics ?
Pages http://localhost:9101/metrics gitlab_pages_domains_updated_total

/cc: @ayufan, @bjk-gitlab, @nick.thomas, @_stark, @dblessing, @jacobvosmaer-gitlab, @zj

Edited Sep 06, 2018 by Ben Kochie