For example:
https://koji.fedoraproject.org/koji/taskinfo?taskID=34567491 https://koji.fedoraproject.org/koji/taskinfo?taskID=34567975
Ignore what koji says, the interesting part of the logs is in root.log:
DEBUG util.py:585: Updating / installing... DEBUG util.py:585: eclipse-1:4.11-7.module_f29+4132+22d25######################################## DEBUG util.py:583: BUILDSTDERR: error: unpacking of archive failed on file /builddir/build/SOURCES/eclipse-platform-sources-4.11RC2.tar.xz;5cca4d1d: cpio: read failed - Inappropriate ioctl for device DEBUG util.py:583: BUILDSTDERR: error: /builddir/build/originals/eclipse-4.11-7.module_f29+4132+22d25fa2.src.rpm cannot be installed
Metadata Update from @bowlofeggs: - Issue priority set to: Waiting on Assignee (was: Needs Review) - Issue tagged with: koji
Yeah, we have seen this before, but are not sure whats causing it.
It seems to be when it downloads the src.rpm from the hub it's somehow corrupt, but yet it doesn't error out until it tries to unpack it. I'm not sure what else to do to track it down, we have been doing various things to reduce load on s390x builders hoping it was load related...
What about some kind of mitigation? Can it be taught to retry if that happens?
(Eclipse is a loooong build to wait for notification of a failure on a single architecture and MBS does not allow me to build with the koji option "--fail-fast")
I've opened https://pagure.io/koji/issue/1418 on this issue for upstream to comment on. Hopefully they can figure out some way to mitigate this.
I'm going to close this in favor of that one... lets follow along there and see what they can do.
Sorry for all this hassle.
Metadata Update from @kevin: - Issue close_status updated to: Upstream - Issue status updated to: Closed (was: Open)
Login to comment on this ticket.