PVE host pauses when network connection has a problem.

[shubes]

I wouldn't have thought this was possible. We're in the process of determining the exact problem with the external network device, but I thought I'd go ahead and see what the experts have to say about this.

I have a PVE 2.0.3 host, with an Intel (82571EB) PCIe dual nic card. One of the ports is attached to an Adtran network device (I don't know which model), which as a T1 coming in to it. The Adtran device appears to have an intermittent problem, stalling for 3-5 minutes at a time, once per day or so at random times. When the Adtran device stalls (phone lines are unresponsive as well), the PVE host pauses as if someone pressed the pause button on a keyboard (there is no KVM on this host). When the Adtran begins working again, the PVE host and all VMs resume w/out missing a step. After the pause event, the performance summary charts in PVE contain a gap (no graph lines at all) that corresponds to the outage period.

What I don't understand is how a network device connection can cause the PVE host to freeze like this. I would expect (and desire) a network failure event to occur instead of the entire server seizing up.

Can someone explain this?

My apologies if this is construed to be off topic in some way. I realize that the problem may not be specifically PVE related.
Thanks.

 

[dietmar]

Never saw such behavior, sorry.

 

[shubes]

Boy, do I feel stupid. After checking the logs, it turned out that the power had failed on the outlet where the UPS was plugged in, and the UPS failed (battery dead). In short, the server (and the network device) lost power, then gracefully came back up when power was restored. Somehow I assumed that the server was up the whole time!

Thanks Dietmar. Sorry for the 'noise'.

 

[dietmar]

In short, the server (and the network device) lost power, then gracefully came back up when power was restored. Somehow I assumed that the server was up the whole time!

yes, that sounds more reasonable ;-)