I setup the vHMC week before last and last week @arrfab got the configuration sorted out.

Due to the way things work, we need to get some more small disks so we can use them for the VIOS (virtual IO server) setup. Otherwise this small instance takes over 2 of the 4 6TB drives and gives us much less room. ;(

Those are in process of being ordered now.
Once those arrive we also need to add 2 more network interfaces (because we want 2 for vnic and 2 for sr-iov network). Also, the 2 existing ones are 10G, when the cards support 25G, so we should also fix that when we get them to hook up the other two interfaces.

So, waiting on the hardware at this point. Once thats done we can migrate some buildvm's to a new lpar on the reconfigured one, then reconfigure the other one.

Tons of updates here. We got the hardware. We got it installed and network configured.

We have run into fun with sr-iov and virtual networking and bridges, we are working through some way to make it all work as we need.

I'm going to work on setting up lpars next week I hope (I am out tomorrow and monday)

For coreos folks:

The network is going to be a bit odd I'm afraid. You will have 2 network devices, which you can bond, but you can't put a bridge on it, only macvtap. Hopefully thats doable... also will need to tag a particular vlan on your side. Once we sort that out hopefully everything else will work.

CC: @dustymabe @poorna

ok. lpars are setup now, we just need to install things.

There's a lpar for fedora coreos
one for bvmhost-p10-02 (builders in prod)
and one for buildhw-p10-02 (a staging builder)

The network is going to be a bit odd I'm afraid. You will have 2 network devices, which you can bond, but you can't put a bridge on it, only macvtap. Hopefully thats doable... also will need to tag a particular vlan on your side. Once we sort that out hopefully everything else will work.

Can we just use one of the devices?

I'm happy to bond them too, just trying to get clarity about the limitations.

yes, but if/when they do switch maint and reboot one 'side' of switches the machine will be not reachable.

They setup things in rdu3 so they could do updates/reboots without affecting connectivity for us, so every machine has a pair of network ports each on a different switch.

Sorry it's so complicated... would have been nice to just give you one virtual nic, but thats not possible.

The first power10 is all setup and has lpars and running fine.

The second one we reconfigured today, but there's something amiss with the switch port config at least, so we need that fixed before we can use it.

So, everything is now reconfigured.

I am going to close this issue as it was around the configuration/setup.

I have filed: https://pagure.io/fedora-infrastructure/issue/12914 to track installing the fcos lpars. :)