[FEATURE] Improve Shipwright Builds Metrics

[avinal]

Is there an existing feature request for this?

• I have searched the existing feature requests

Is your feature request related to a problem or use-case? Please describe.

Currently, Builds exposes several basic metrics. Here are the existing metrics:

Metric	Type	What It Tracks
build_builds_registered_total	Counter	Builds that pass validation
build_buildruns_completed_total	Counter	BuildRuns where an executor was created (misleading name)
build_buildrun_establish_duration_seconds	Histogram	Time from BuildRun creation to executor start
build_buildrun_completion_duration_seconds	Histogram	Time from BuildRun creation to executor completion
build_buildrun_rampup_duration_seconds	Histogram	Time from BuildRun creation to executor creation
build_buildrun_taskrun_rampup_duration_seconds	Histogram	Time from executor creation to pod creation
build_buildrun_taskrun_pod_rampup_duration_seconds	Histogram	Time from pod creation to last init container finish

With enabled labels, they can track plenty of stats, but some use cases remain undiscovered. Which I think will be useful to the user.

Describe the solution that you would like.

I am proposing these enhancements to Builds metrics.

Missing labels (not available on any metric, even as opt-in)

Label	Values	Why it's needed
strategy_kind	BuildStrategy, ClusterBuildStrategy	Distinguish namespace-scoped (custom) vs cluster-scoped (Red Hat provided) strategies. Currently impossible to tell.
executor	TaskRun, PipelineRun	Distinguish executor type. PipelineRun support was recently added but no metric tracks which executor ran a build.
result	succeeded, failed, cancelled, timeout	Build outcome. No metric tracks success vs failure today. This is the single biggest gap.
source_type	Git, OCI, Local	Source code delivery method. No metric tracks this.

Missing metrics

Metric	Type	What it would track
build_buildrun_result_total	Counter	BuildRun completions with outcome, the metric that build_buildruns_completed_total should have been
build_buildrun_failure_reason_total	Counter	Failed BuildRuns broken down by failure reason (OOM, eviction, timeout, etc.)
build_buildruns_active	Gauge	Point-in-time count of currently running BuildRuns
build_buildstrategy_count	Gauge	Point-in-time count of BuildStrategy and ClusterBuildStrategy objects

Existing metric issues

Metric	Issue
build_buildruns_completed_total	Misleadingly named, incremented when executor is created, not when it completes. Also has no outcome distinction. It should be deprecated in favor of build_buildrun_result_total.
build_builds_registered_total	Only incremented on successful validation. Builds that fail registration (e.g., strategy not found) are invisible.

Describe alternatives you have considered.

No response

Anything else?

No response