action #98496
closed
nightly circle CI job fails in "Install cached packages" with "rpm: no packages given for install" size:M
Added by okurz over 2 years ago.
Updated over 2 years ago.
Description
Observation¶
https://app.circleci.com/pipelines/github/os-autoinst/openQA/7659/workflows/cc8ba326-aa47-40a4-b43c-fec12d1704b1/jobs/72308 shows
#!/bin/bash -eo pipefail
set -x
if [ -z "$CIRCLE_WORKFLOW_ID" ]; then
bash .circleci/build_cache.sh
else
sudo rpm -i -f $(find /var/cache/zypp/packages/ | grep '.rpm$')
fi
+ '[' -z cc8ba326-aa47-40a4-b43c-fec12d1704b1 ']'
++ find /var/cache/zypp/packages/
++ grep '.rpm$'
+ sudo rpm -i -f
rpm: no packages given for install
Exited with code exit status 1
CircleCI received exit code 1
in a previous step "Restoring cache" there is:
No cache is found for key: v1-R3CxihB_CRf9AaxlDij+JU8Ke6fsRQrgKsZfJNeqcCs=-0ctxsgSE3h8sZFrULdZqhQKTjwrMfWf9GrviDcdOQG8=
Acceptance criteria¶
- AC1: CI should fail w/ a clear error message if the cache is empty
- Subject changed from nightly circle CI job fails in "Install cached packages" with "rpm: no packages given for install" to nightly circle CI job fails in "Install cached packages" with "rpm: no packages given for install" size:M
- Description updated (diff)
- Status changed from New to Workable
- Assignee set to Xiaojing_liu
- Description updated (diff)
Updated the AC after discussing it with Jane and Oli
We discussed about this and found that my initial request was no good. We use the cached packages also to define our consistent, version specific set of packages that we want to rely on. Resorting to a generic zypper call on errors might work but defeat the purpose as we would not realize if there are problems on saving and loading the cache so eventually we would end up always downloading different versions with zypper on the fly which we wanted to avoid. For now I don't have any better idea than https://github.com/os-autoinst/openQA/pull/4238 for now, it's really not much.
Our next best bet would be to find the job on circleCI that actually failed to properly save the cache and prevent the error the next time.
- Assignee deleted (
Xiaojing_liu)
I have no idea how to find the job that failed to cache. Searched the CI jobs and found nothing. I release this ticket.
- Status changed from Workable to Rejected
- Assignee set to okurz
somehow this hasn't happened anymore. So something might have fixed it, maybe it will come back.
Also available in: Atom
PDF