VMs temporarily malformed after restore

[LooneyTunes]

This is as strange as it sounds. If anyone has a good explanation I'm all ears.

My disk crashed, and now got a new fresh 7.0 installation running. The strange part... I restored my VMs and started them. After a running for a while, they suddenly went belly up. I could not even access the shell within proxmox... I then went into the respective VM's hardware setup and found that not much was the way it should. It looked like some spooky default values at best. A restart "fixed" that issue, but really...?

 

[fabian]

sounds like the GUI might have displayed some default values for fields for some reason? if it happens again, can you happen the browser dev tools and check the console there for errors?

 

[LooneyTunes]

sounds like the GUI might have displayed some default values for fields for some reason? if it happens again, can you happen the browser dev tools and check the console there for errors?

Hi,
it was more than GUI issues. The machines borked as well, ie mallfunctioning, seemingly reconfigured, that is how I noticed it. I'd be happy to get you any errors. But I can't see how a mere webbrowser can debug Proxmox? Can you please elaborate a little of what you would need?

 

[fabian]

well.. without any more details it will be hard to debug check the logs from around the time the problem happened (both host and guest), maybe they contain some pointer.

 

[LooneyTunes]

well.. without any more details it will be hard to debug check the logs from around the time the problem happened (both host and guest), maybe they contain some pointer.

fully agree on that one. Will check the logs for something to back this up. Well... Can't find anything in the logs to support this, so guess that's it for now.

 

[LooneyTunes]

I got this again just now. And again, nothing made any sense, I could not access anything in that state, but took some screenshots. It seems to have been some login-token(?) having expired, as after a while the login dialog was shown, and after login things were back to normal.

I found this in the syslog, corresponding to the event

Aug 14 12:03:23 pve pveproxy[39321]: problem with client ::ffff:192.168.1.2; Broken pipe
Aug 14 12:03:24 pve pveproxy[39321]: Can't call method "push_write" on an undefined value at /usr/share/perl5/PVE/APIServer/AnyEvent.pm line 359.
Aug 14 12:03:29 pve pveproxy[39321]: worker exit
Aug 14 12:03:29 pve pveproxy[959]: worker 39321 finished
Aug 14 12:03:29 pve pveproxy[959]: starting 1 worker(s)
Aug 14 12:03:29 pve pveproxy[959]: worker 46784 started
Aug 14 12:08:14 pve pvedaemon[953]: <root@pam> successful auth for user 'root@pam'

 

[RolandK]

i also see this in log :

Can't call method "push_write" on an undefined value at /usr/share/perl5/PVE/APIServer/AnyEvent.pm line 359.

 

[fabian]

that should be harmless, although it could be improved. if the connection goes away client-side at the right moment, the server attempts to but is not able to send a reply, that triggers this log message. or do you also see this in connection with other issues?

https://git.proxmox.com/?p=pve-http...bd2a4c8944e108108a4;hb=refs/heads/master#l359

see the 'return if !..' in line 364 in the other branch that was introduced for a similar spurious warning..