jenkinsci/prometheus-plugin 873.v054598e6e84d on GitHub

What's Changed

The plugin no longer depends on pipeline-stage-view (pipeline-rest-api, com.cloudbees.workflow.rest.external.*). It now uses workflow-api, workflow-job and workflow-step-api directly.
Stage metrics (stage_duration_milliseconds_summary, stage_result_ordinal) are now computed by the plugin itself from the flow graph.
stage_result_ordinal values are unchanged. The order of the status enum is documented as part of the metric contract.
Interrupted stages now use the result of the FlowInterruptedException (e.g. ABORTED, UNSTABLE).
Stages skipped by Declarative Pipeline (when, earlier failure, restart) are reported as NOT_EXECUTED with duration 0.
Stages using catchError, unstable or warnError are reported according to their worst recorded result.
Stage IDs are now sorted numerically, so 10 comes after 9.

Fixes a deadlock on legacy Jenkins cores. The build count now uses an approximation based on getNextBuildNumber().
The warning about the missing CloudBees Disk Usage Simple plugin is now logged only once instead of on every scrape.

pipeline-stage-view is no longer pulled in transitively. If you need the Stage View UI, install it separately.
Stage values can differ slightly from before in edge cases (skipped, aborted and unstable stages, nested or parallel stages). Please check dashboards and alerts based on stage_result_ordinal and the stage summary after upgrading.
The build count in NbBuildsGauge is now an approximation and may include deleted or skipped build numbers.