I recently started setting up some LXCs that use Option 3 from this thread to share network storage.
[Tutorial] Mounting NFS share to an unprivileged LXC
Last night one of the LXCs was doing a bunch of disk IO, and I tried to run the "pct set 103 -mp0 /host/dir,mp=/container/mount/point" command from the host Shell via the web UI and the command hung. At that point I wasn't able to access the NAS either. I checked the console for the NAS VM, and it was in some sort of loop. I took a video and will try to post it.
I restarted the NAS VM and all was well, until today when the whole proxmox system seemed to soft crash. Web UI was showing updated IO Delay between 13% and 26%, but the LXC's consoles wouldn't open, status' weren't updated etc.
I couldn't get the VMs to shutdown, or host to restart. I had to go into the IPMI to shut the power off and reboot. Things are working again now.
More info. The NAS is TrueNAS Scale and is a VM on the host. The drives are all passed through to the VM from the controller card they're attached to. This has been working solidly for over a year.
Questions.
1. Is this Option 3 method the best way to get a NAS share into an unprivileged LXC?
2. Is the data cached on the host drive where the map point is mounted? Is the physical drive a limitation here?
3. The connection from TrueNAS to the host is via NFS and the network, could the NIC be an issue?
4. In the attached screenshot you'll also see that the SWAP usage is quite high. I've never seen it above 10% before.
5. Is there anything that typically causes a host to become unresponsive like this?
Please let me know what other info I can provide.
[Tutorial] Mounting NFS share to an unprivileged LXC
Last night one of the LXCs was doing a bunch of disk IO, and I tried to run the "pct set 103 -mp0 /host/dir,mp=/container/mount/point" command from the host Shell via the web UI and the command hung. At that point I wasn't able to access the NAS either. I checked the console for the NAS VM, and it was in some sort of loop. I took a video and will try to post it.
I restarted the NAS VM and all was well, until today when the whole proxmox system seemed to soft crash. Web UI was showing updated IO Delay between 13% and 26%, but the LXC's consoles wouldn't open, status' weren't updated etc.
I couldn't get the VMs to shutdown, or host to restart. I had to go into the IPMI to shut the power off and reboot. Things are working again now.
More info. The NAS is TrueNAS Scale and is a VM on the host. The drives are all passed through to the VM from the controller card they're attached to. This has been working solidly for over a year.
Questions.
1. Is this Option 3 method the best way to get a NAS share into an unprivileged LXC?
2. Is the data cached on the host drive where the map point is mounted? Is the physical drive a limitation here?
3. The connection from TrueNAS to the host is via NFS and the network, could the NIC be an issue?
4. In the attached screenshot you'll also see that the SWAP usage is quite high. I've never seen it above 10% before.
5. Is there anything that typically causes a host to become unresponsive like this?
Please let me know what other info I can provide.
Attachments
Last edited: