On Fri, 27 Jan 2017 09:18:30 -0600 Michael Cronenworth <mike@xxxxxxxxxx> wrote: > On 01/26/2017 09:13 PM, Kevin Fenzi wrote: > > f25 updates testing you mean? > > > > It's the atomic part of the compose failing. Folks are working on > > it. > > > > It's being tracked in: > > https://pagure.io/releng/issue/6602 > > I realize that breakage doesn't happen often, but when it does it > more often takes days to repair. The down times I've seen seem to > stem from "rpm-ostree" issues. If we're going to bring in new, > un-tested features (that only bodhi uses) can we get a staging > instance setup for bodhi? That way we can attempt to reduce these > breakages? Any other commercial entity would have the same, tiered > environment setup. We do have a staging bodhi setup. The problem is that the composes are done with the contents of the compose. So, when a new rpm-ostree update is submitted to f25-updates-testing that very new update is used in the compose. I don't think it's practical to first try every compose in staging with an update that could be in the compose. I do think there might be some things we could do better here tho, like if a compose fails, have a way to drop just those few packages that are causing the failure out and repushing the rest until we can sort it out. (and making this the SOP, not just a last resort). Or perhaps have some test/CI of the rpm-ostree set of packages that confirms they work in the env that bodhi uses before they are submitted as updates. In any case, sorry f25-updates-testing is stopped up. People are working on it. Hopefully it will be fixed soon. kevin
Attachment:
pgp0tarzgFhZi.pgp
Description: OpenPGP digital signature
_______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
