This repo contains scripts that Ed uses regularly, and brief descriptions of how/when/why they're used.
Scripts I run when needed:
- ci-watch - console tool for monitoring running CI jobs. I run it pretty much on every PR I submit and resubmit.
- cirrus-vm-get-versions - generate table of package
versions on
automation_images
(new-VM) PRs. I run it every time I get that little table email showing VM 'cYYYYMMDD' numbers. - get-cirrus-buildid-for-pr - one of the most useful scripts I have. Feed it a PR, it will (if possible) spit out the Cirrus BuildID for that PR. (Remember, a Cirrus build is a set of many tasks, each of which is a VM-test combination). I don't actually run it manually, but many of the scripts in this tree call it.
- git-sha-to-pr - given a git SHA, shows its corresponding PR.
- orphan-kill - helper for handling "orphan VM" emails. I run it once a week or so, when I get one of those orphan emails.
- flakes - the big one. Tools for tracking and cataloging flakes.
- plots - generate time-series plots showing how long CI jobs take. This one should run from cron, maybe every 2-4 hours. It's just too new so I wanted to get a sense for it (results, unexpected hiccups) before letting go.
Mostly run from cron, with occasional human intervention when I see unexpected results in my morning email:
- cirrus-credits - track Cirrus credits, warn if low
- cirrus-cronjobs - simplified maintenance of nightly CI runs
- quay-check - make sure quay.io images have reasonable descriptions
This one is a systemd unit, not cron, but same idea, it runs unattended:
- openqa-listener - alert if podman openqa jobs fail.
Hodgepodge of little tools that make my life easier.