This ticket involves nightly composes of the desktop spin:
We, the QA team, use these composes quite frequently for various Test Day events etc, which means it fairly important to us. Unfortunately the current implementation of nightly compose build system always deletes the old ISO prior to building a new one. That means if a new build fails (because of broken dependencies between packages, etc) then there is no ISO available for us to use.
I would like to request a change in the building script that would leave the old ISO intact until the new ISO is successfully built. In that case the old ISO can be deleted. If the new build fails, the old ISO should stay available.
As discussed with Kevin Fenzi he stated the the biggest problem was the lack of disk space. He said that there was not enough space for the old ISO and the new ISO to co-exist. That raises my question - do you, the Infrastructure team, have some solution to this problem? Do you have some spare disks available that could be added to the existing system? Or maybe the whole building process of nightly composes could be transferred to a system with more disk space? Do you need help with finding a budget for the new disk?
Thanks for answers.
I think I have gotten this solved.
In the script, when a compose fails on spin1, I add an exclude for it to an exclude file.
Then, this exclude file is used when rsyncing over, so it will not delete the old iso version thats there.
Its hard to test this out right now, because everything except aos is composing fine and aos is a odd case. ;)
I will leave this ticket open for a bit and make sure my fix works.
Everything should be working on this now. It was a bit of a pain with rsync, but I have it working.
Feel free to re-open or file a new bug if you spot any further issues.
to comment on this ticket.
Copyright © 2014-2017 Red Hat
2.15.1 — Documentation