#11318 ppc64le koji tasks failing with OSError: [Errno 28] No space left on device: '/var/tmp/koji/tasks/*'
Closed: Upstream 2 years ago by kevin. Opened 2 years ago by decathorpe.


Also, looks like the first build has technically "failed" but since the build itself succeeded and only the "tagBuild" subtask failed ... it's now stuck as "failed" but I can't resubmit it: https://koji.fedoraproject.org/koji/buildinfo?buildID=2201477

@decathorpe could you check wich items is taking more space on disk ??

du -ms /var/tmp/koji/tasks/* | sort -hr |more

The problem seemed to be that builders were not able to remove old epel7 buildroots (and since that errored out no build roots were removed).

For example:

May 08 20:02:51 buildvm-ppc64le-28.iad2.fedoraproject.org kojid[3007743]: 2023-05-08 20:02:49,651 [INFO] {3007743} koji.TaskManager:878 Removing buildroot: 42169432/epel7-build/ppc64le
May 08 20:02:51 buildvm-ppc64le-28.iad2.fedoraproject.org kojid[3007743]: 2023-05-08 20:02:49,652 [ERROR] {3007743} koji:517 Unable to remove directory lib: [Errno 39] Directory not empty: 'lib'
May 08 20:02:51 buildvm-ppc64le-28.iad2.fedoraproject.org kojid[3007743]: 2023-05-08 20:02:49,652 [ERROR] {3007743} koji:517 Unable to remove directory var: [Errno 39] Directory not empty: 'var'
May 08 20:02:51 buildvm-ppc64le-28.iad2.fedoraproject.org kojid[3007743]: 2023-05-08 20:02:49,652 [ERROR] {3007743} koji:517 Unable to remove directory root: [Errno 39] Directory not empty: 'root'
May 08 20:02:51 buildvm-ppc64le-28.iad2.fedoraproject.org kojid[3007743]: 2023-05-08 20:02:49,652 [WARNING] {3007743} koji.build:96 file removal failed for /var/lib/mock/epel7-build-42169432-5132696

So, I cleaned manually out all the /var/lib/mock/epel7-build* repos and things mostly got back to working.
However, there was one more machine that was saying it was out of space, even when it had a bunch of space available.
I rebooted that one and it didn't come back, so I am just going to reinstall it. It must have become corrupted somehow.

I went and retagged all the things that had failed tagbuilds.

I'll leave this open however until I can file an upstream koji ticket on why epel7 buildroots aren't getting cleaned up as expected (ie, the underlying cause).

Metadata Update from @zlopez:
- Issue priority set to: Waiting on Assignee (was: Needs Review)
- Issue tagged with: koji, low-trouble, medium-gain, ops

2 years ago

Metadata Update from @kevin:
- Issue close_status updated to: Upstream
- Issue status updated to: Closed (was: Open)

2 years ago

Log in to comment on this ticket.

Metadata
Boards 1
ops Status: Backlog