> I'm the secondary architecture release engineering lead.
I've not yet
> decided how to proceed with s390 support in EPEL. I've got a couple of
> ideas but nothing that is final. In the next few weeks we'll be
> building and importing ppc64le and once that is complete, as it's the
> first new EPEL arch added in some time, I'll have a better idea how
> best to proceed.
>
> I've got grave concerns about remote builder capability. We would need
> to use the internal Red Hat mainframe but the remote ppc64le builders
> for COPR have been less that stellar causing consistent issues for the
> infrastructure team so to say I'm cautious in this regard doesn't even
> come close. I have another idea but I need to find some time to test
> that option to see if it's feasible.
Hi Peter, thanks for the update. Does the internal Red Hat mainframe
count as a remote builder as well? Presumably it wouldn't have the
same reliability issues that plague the remote ppc64le builders. For
COPR are you looking for additional hardware or will you use the internal
Red Hat mainframe as well?
If it's not in the same physical datacentre it's classed as remote. So
yes, it's remote. I can't say anymore about Red Hat's infrastructure.
The COPR infrastructure is a completely unrelated set of
infrastructure, it was used as an example of how problematic remote
infrastructure can be.
These are all considered production because a lot of people rely on
it. People get woken up in the night and on the weekends if things are
unavailable. This is a problem for me even if it's not me that's being
woken up, we need to ensure there's escalation paths, means of
mitigation etc. I'm not currently got time to deal with that. It's on
my todo list, no I don't have a ETA.
Peter