In order to detect if a certain build host is causing flakiness there should be a dashboard that tracks failed ci jobs per build host. If hosts are operating normally the failed jobs should distribute evenly among hosts.

Fails should be filterable by

• branch
• project
• workitem type (build or test)

Each HW pool needs a own section

• Windows/Linux hosts (kvm)
• Intel Mac minis (parallels)
• ARM Mac minis (jenkins)