So I tried adding a bonded pair using my 2 10Gig interfaces, one of which was my management interface… needless to say I lost connectivity. So I log into my master, reset the management, and my servers start to come back. Except, on two boxes, when I go to reset the network interface, it now says there is NO network interfaces at all. I tried doing an emergency network recovery, and nope… nada. Still no interfaces.
If I try almost any network related CLI commands they just sit there and do nothing.
Is there a way to reset a host from the CLI back to “default” without having to re-install ? (I didn’t yet take a backup of the server config itself).
I have never had it not work when using the xoconsole, but’s it’s been a long while since I broke a system.
I ran into a similar problem of disappearing NICs. Thankfully it wasn’t my management interface so I can still access the box. I can’t seem to get my 10 gig nics added back to the box. Emergency networking reset doesn’t work, no scanning from the console works. Nic shows up properly in lspci though.
Let me know if you figure out how to get them back please. I’d like to give your solution a try on my setup.
Unfortunately, this was a new setup for me, and it was faster to simply re-install XCP-NG (I had no VMs on it).
But if you still have the management interface, move your VM’s off, unjoin the pool, re-install, re-join. Should take ~1hr.
Thanks for the hint. I only have a single host to deal with so I can’t move the VMs off unfortunately. I might just shut it down and do a reinstall on another drive - finally retire the spinning rust in that machine.
It’s just a home lab box so a little bit of downtime isn’t the end of the world. I’ll see about the process of doing a fresh install and importing the VMs into it this weekend sometime.
I’ve posted a few different scenarios that seem to cause this, and it seems as though, the real solution IS to re-install the OS. xapi gets confused with certain nic configs, also if your NIC is broadcom based, they are known to not be good, and experience random disconnects, complete.
Well I figured out my issue. After I reinstalled and managed to nuke my virtual machine storage in the process. Thankfully i was able to restore from backup.
Anyways I was just SSHing into the console of the machine as it runs headless. I noticed while it was rebooting on my reinstall that it was complaining about unsupported SFPs in that nic. I was using FS.com SFPs and not genuine Intel ones.
Anyways I retired my spinning rust storage as well in the process. XCP-NG is now installed on a 240 GB SSD and my virtual machine storage is on a 1 TB NVME drive.
I found some Intel SFPs on ebay and they should be here by the end of the week so i should be all set after that.
1 Like
Most off-brand SFP’s are reported as unsupported. I highly doubt those are the source of your problem. What I have been told, is that broadcom based NICs are super unreliable in XCP-NG (and Xenserver as well) and to basically get rid of them.
As it turns out, it is well know that changing a management interface can royally screw up xapi’s ability to sort out network interfaces. The recommended fix is to actually re-install the host.