Hide Forgot
In case the rpm-test-trigger service is down (e.g. after crash) for a certain period of time (e.g. 60s to hours), we'll need a strategy to find advisories which changed state during the down time to scheduled tests.
We need structured results implemented by the Errata team to implement this. See Bug 1342878
We had a discussion about this to use durable queues. But we came to the conclusion that this might not be a reliable solution, since message queues might fail in times of down time as well. Perhaps it can be as simple as: query ET for new items without a test result recorded.
Anyway, no matter if https://git.beaker-project.org/cgit/rpm-test-trigger/ is going to consume errata or builds, it still needs to have some stored value which would say what errata/builds were already processed or what was the last time it got any command right? (I'm just trying to figure out what is the objective here, sorry for dull questions) In case of Koji/Brew, we might save taskID? $ koji list-tasks --channel default ID Pri Owner State Arch Name 21083515 20 ellert OPEN noarch build (rawhide, root-6.10.04-1.fc27.src.rpm) 21083520 19 ellert OPEN aarch64 +buildArch (root-6.10.04-1.fc27.src.rpm, aarch64) 21083522 19 ellert OPEN armhfp +buildArch (root-6.10.04-1.fc27.src.rpm, armv7hl) 21083524 19 ellert OPEN i386 +buildArch (root-6.10.04-1.fc27.src.rpm, i686) 21085114 20 rjones OPEN noarch build (f27-ocaml2, /rpms/ocaml:59545eed91afae7ab19d8010e3ea964372653614) 21085164 19 rjones OPEN aarch64 +buildArch (ocaml-4.05.0-2.fc27.src.rpm, aarch64) [...]