Host crash brings down network
-
I have been using XCP-NG for a home lab and experimenting with different server software and have gotten by with very little trouble, but now I have an issue and I don't even know where to start trying to resolve it.
My host is an older computer I built in 2015, after building a new PC I put xcp-ng on it to play around.
i7-4790k
16gb ram
nvme drive and a few ssds
an old amd tahiti 7970 that is not being passed through to anything, yetit is connected to a switch that has my other computer attached, which goes to another switch, which leads to the router and ends at the modem.
I have found that occasionally while trying to start a newly made windows 10 vm that I was going to pass the gpu through to, starting windows 10 will break everything.
I found that the order things seem to happen is :
I start VM
VM looks like it is going to boot, but then the console view in xoa stops working
xoa disconnects
~15 seconds later all network activity on any device connected by ethernet stops working. But wifi still works.All network activity over ethernet will stay down until I hold the power button down on my host to kill it and bring it back up.
When this has happened I am unable to ssh anything or use xoa for anything.
Any suggestions on how to figure out what is happening?
-
A few more details,
This happens about 50% of the time while trying to start the windows 10 vm.
It is only this VM, and I have tried deleting it and making another windows 10 with the same result.
the ethernet network comes up immediately after killing the host. by running a continuous ping I can restore the network after a single missed packet if I am quick enough.
there are about 5 pcs plugged in to ethernet that this effects
-
I would start by checking the logs -- https://xcp-ng.org/docs/troubleshooting.html
How is networking configured on the problematic VM (ie: static IP vs DHCP). From you description, it sounds like you may have an IP address conflict.
-
@danp The vm is set to dhcp with the virtual nic pulling an IP from a pfsense vm I am running on the same host.
I do have a mac reservation set for the host, xoa, and pfsense
-
After removing the GPU I am unable to replicate the issue.
I have put the GPU back in and still cannot get the issue to happen again.
Issue must have been some hardware related issue, but I'm still at a loss as to how things happened the way they did.