#9059 copr: ppc64le HW machines re-installed in new lab?
Closed: Fixed 2 years ago by praiskup. Opened 3 years ago by praiskup.

We are keeping repeated questions about this topic, so I'm going to
filll a new tracker issue for this topic (we touched this a bit in another issue before).

ETA for this is mid-August. At the moment there are too many issues with one datacentre to look at it before then.

Thank you for the update!

Hm, OK, that's a pity. We basically don't need
anything else than rack the boxes, and have some EL7+ system installed on
the boxes (+ some range of IP addresses). So if we could help you somehow to
make it happen sooner, please let us know.

The problems are that is not that simple. You need to have a working switch set, you need a working router, and you need some auxiliary services working in that datacentre. The datacentre is under quarantine which means we have to plan a week or so in advance to have people do this work. Due to the way that the community cage is set up, there is also only 1 person in the company who can do most of the work from the network and installation side, and they are working this month and the next on getting the main Fedora infrastructure working in IAD2. When that work is done they will be taking a breather and then RDU-CC can be worked on.

No doubt, this isn't an easy time. Thank you for your work.

Metadata Update from @smooge:
- Issue priority set to: Waiting on Assignee (was: Needs Review)
- Issue tagged with: cloud, groomed, high-trouble, medium-gain

3 years ago

Per recent discussion on fedora-admin channel:

<sharkcz> nirik: smooge: hi, do you have an estimate when ppc64le hw will be available in the new DC for copr? seems many people are missing it :-)
<smooge> no
* sdoherty has quit (Ping timeout: 260 seconds)
<smooge> I wish I had a better answer like as soon as we can.. but that is just a cleaned up 'no I do not have an estimate

Any estimation would be nice, there are now even other requests:

I just want to raise awareness that there are multiple factors to how we can service this which is why we can't give a firm estimate. There is a singular network administrator which is not part of our team and whom we rely on to get work like this done. That admin is working full time on another data center and several other items so getting that persons time is difficult to plan for, especially when this is a community cage not under his remit. Smooge is also the only sys admin who can take this task on and is currently at max capacity due to the colo move and other higher priority initiatives. We need both of these to align at some point in the future for us to get this working and we have empathy with the situation you are in but would request your patience. We are not forgetting about this, we just can't service it right now. Please reach out to me directly if you have concerns about the timeline.

Metadata Update from @smooge:
- Issue tagged with: ops

2 years ago

I understand that the team is overloaded, but the missing hardware would bring
high gain to copr build system. Users are missing the hardware, it is a regression.

It would be nice to see where the problem is, to somehow define the steps that
need to be done, and delegate as much as possible to our team.

Hi, I will be working with @dkirwan on getting this set up. I have started an internal document to cover what is required, wanted, and what can be delivered when. We will be moving this up the priority queue

Metadata Update from @smooge:
- Issue assigned to smooge

2 years ago

Are these machines POWER8 or POWER9? Thanks.

The systems are power8.

One machine is now up and ready for use: vmhost-p08-copr01.rdu-cc.fedoraproject.org.

The other(s) are gonna take some more work ;(

Both Power8 machines are now installed and working (28 builders). Thank you!

Metadata Update from @praiskup:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

2 years ago

Thanks to everyone involved :-) It will be a great help for various projects.

Login to comment on this ticket.

Boards 1
ops Status: Done