Hello,
I have a pve node on a AMD machine (X570s + 5600).
I have a raidz on the host which the guests (ubuntu containers and the ocassional vm) acess via NFS mount controlled with autofs.
All containers share an internal network via a bridge not attached to any nic on the node for the nfs mount, as well as a secondary bridge on a nic for internet acess
Due to rising energy costs, I run this node o schedule via a pikvm machine that contols the atx pins on the motherboard.
Out of the containers on the node, 2 hang once the shutdown is requested: a qbittorent lxc and one that runs sonarr/radarr/jackett
Looking at the shutdown output, it process hangs either o “pve-guests” service or specifically on the 2 nodes above.
If before shutting down the node, I ssh into those 2 containers and shut them down by hand (eg “poweroff”), and only then initiate the shutdown for the node, everything goes down smoothly.
So it seems the shutdown from the system / node requests shutdown from guest containers in a way that results in a hang.
How should I go about debugging this and figuring out what’s going on?
I am actually considering creating a hacky flow on cron that shuts down those two containers 3 minutes before the actual node shuts down.
But I would like to understand what’s going on as well
.
PS: Errors:
10.10.10.51 is the IP of the host, so it looks like when the shutdown is initiated, the nfs server on the hots goes kabluie and some of the guests are left... hangin'?
Any ideas appreciated!
I have a pve node on a AMD machine (X570s + 5600).
I have a raidz on the host which the guests (ubuntu containers and the ocassional vm) acess via NFS mount controlled with autofs.
All containers share an internal network via a bridge not attached to any nic on the node for the nfs mount, as well as a secondary bridge on a nic for internet acess
Due to rising energy costs, I run this node o schedule via a pikvm machine that contols the atx pins on the motherboard.
Out of the containers on the node, 2 hang once the shutdown is requested: a qbittorent lxc and one that runs sonarr/radarr/jackett
Looking at the shutdown output, it process hangs either o “pve-guests” service or specifically on the 2 nodes above.
If before shutting down the node, I ssh into those 2 containers and shut them down by hand (eg “poweroff”), and only then initiate the shutdown for the node, everything goes down smoothly.
So it seems the shutdown from the system / node requests shutdown from guest containers in a way that results in a hang.
How should I go about debugging this and figuring out what’s going on?
I am actually considering creating a hacky flow on cron that shuts down those two containers 3 minutes before the actual node shuts down.
But I would like to understand what’s going on as well
.PS: Errors:
Code:
Job pve-guests.service/stop running (2min / no limit)
nfs: server 10.10.10.51 not responding, still trying
nfs: server 10.10.10.51 not responding, timed out10.10.10.51 is the IP of the host, so it looks like when the shutdown is initiated, the nfs server on the hots goes kabluie and some of the guests are left... hangin'?
Any ideas appreciated!
Last edited: