Add a multiple prometheus metrics:
Add a multiple prometheus metrics:
Metrics from Machine Provider (Auto-scaling):
- ci_machines_provider{state=acquired|creating|idle|used|removing} - current number of machines in given state,
- ci_machines{type=created|used|removed} - total number of machines that were created, used (number of builds) and removed,
Metrics from Builds with breakdown on stage and runtime state:
- ci_runner_builds{state=pending|running|finished|canceled|terminated|timedout, stage=prepare_script|build_script|after_script|archive_cache|upload_artifacts} - current number of builds in given state
This should basically help us to detect any anomalies in runner: ex. a high number of canceled builds or large amount of machines being provisioned and not removed.
@tmaczukin Can we ship that with 1.9.0RC1?
cc @pcarranza