Add runner_name label to runner-specific Prometheus metrics
What does this MR do?
This MR adds the runner_name
label to all runner-specific metrics:
gitlab_runner_request_concurrency_exceeded_total
gitlab_runner_request_concurrency
gitlab_runner_limit
gitlab_runner_jobs
gitlab_runner_jobs_total
gitlab_runner_job_duration_seconds
gitlab_runner_failed_jobs_total
gitlab_runnerapi_request_statuses_total
Why was this MR needed?
In various setups there are multiple runner instances active on the same host. Right now the metrics only include the runner's token (in a short version) as label which makes distinguishing metrics cumbersome.
What's the best way to test this MR?
Execute various runner operations and then check the metrics exposed on {listen_address}/metrics
. Besides the runner
label there should now also be a runner_name
.
What are the relevant issue numbers?
Backwards compatibility
Since this adds a new partition of the mentioned metrics, users with longer runner metric retention might need to modify their queries in order to get aggregated rules for metrics with and without the new label.
Edited by Horst Gutmann