Learn more about these different git repos.
Other Git URLs
placeholder #3276
Metadata Update from @jobrauer: - Custom field Size adjusted to None - Issue tagged with: testing-done
Commit 1728c07 fixes this issue
It creates problems with image builds. I'm going to remove it from current release and look at it again in 1.30.
Metadata Update from @tkopecek: - Issue status updated to: Open (was: Closed)
Metadata Update from @tkopecek: - Issue untagged with: testing-done, testing-ready - Issue set to the milestone: 1.30 (was: 1.29)
https://pagure.io/koji/c/5b8465a0e817c2e130e8298c05af8bca6a27ccba?branch=master
@jcpunk Have you found what exactly is happening that PrivateTmp help with for kojid? I'm not sure if I can see that. We've hit a few problems (especially imagefactory's access to kojid files, etc.), so we can't enable this behaviour by default as it block non-rpm tasks.
For kojira it still should be ok to enable it.
I've be experimenting with some tooling to spin up additional koji builders under high load. My spin down code is a bit flaky. When there is an unclean shutdown under the non-private temp, when kojid comes back up it gets very confused by the stale files. If I've made a sufficient mess, kojid wont accept new jobs until I clean up those files as it thinks there are jobs running.
Hmm, found the culprit: https://github.com/libguestfs/libguestfs/blob/7ed0da779fd040e0b14414e9a26bfb879ec5fab4/lib/tmpdirs.c#L166 hardcodes /tmp for socket communication for root processes. And it is in different namespace in that moment. So, it looks like no-go for kojid. For kojira it is ok - nevertheless, no part of kojira is using tempdirs/files now.
Metadata Update from @tkopecek: - Issue close_status updated to: Dropped - Issue status updated to: Closed (was: Open)
Metadata Update from @tkopecek: - Issue set to the milestone: None (was: 1.30)
Login to comment on this ticket.