From: Justin M. Forbes on gitlab.com
1. Right, the pipelines fail, I have to manually fix this stuff up every time
a new one appears. When it is fixed, there is no automated process to go and
re-run failed pipelines. I do not usually rerun the pipelines, typically
Patrick does, and not on weekends/holidays. As you can see, I have rerun
!1832 pipeline for you and it passes now. It would have done the same
2. I was not aware of that, it seems that new entries of zfcpdump are only
added when they differ from s390x as there are about twice as many config
entries for s390x compared to s390x/zfcpdump. If anyone wanted to change
s390x back to =m for any reason, the config would fail and the MR would never
get through without adding back an entry for zfcpdump, but as the entire
flavor is a RHEL specific thing, I am happy to do whatever RHEL likes.
3. I do not disagree, it should be cleaned up. My point was ark is a fine
location, as Fedora does not build those configs. Common is for things which
are identical between Fedora and RHEL/ARK. The ark directory is for things
that are only set that way in RHEL/ARK. There are probably a large number of
configs that need to be cleaned up in a similar manner because for the first
year+ of this workflow, all new RHEL entries were created in common instead of
ark. That only changed a few months ago.
On weekends and holidays, I only check on things enough to make sure that they
are not failing the merge or failing the configs so that people can continue
to work if they wish to do so. A RHEL mismatch cleanup means I have to create
an MR after I fix it in pending so it does not get lost or forgotten.