Does anyone have a working example of a host virtiofsd command line, and a VMID.conf arg: line, that works with the "new" virtiofsd in proxmox 8?
I tried the arg: line I used from a pve7.4 VM (which worked fine once virtiofsd was running in 7.x) but it does not appear to work with version 8.
I see the example here, but once I put the qemu-system arguments in the args: line, virtiofsd starts in my hookscript, then dies. If I add --debug-level trace, no additional or relevant data is added that helps troubleshoot. It just dies. I've tried so many variations I've lost track... none of them have worked.
I currently have the following in my VMID.conf file:
And run the following from my hookscript in the pre-start phase of the hookscript:
All I get in my virtiofs-start.out file is:
Before it dies on qmstart with no further output.
If I restart virtiofsd by hand, which appears to stay running using the same command line but started manually, then try to mount a directory in a guest (Ubuntu 23.04) with:
But then:
ls /mnt/test/rcc
will hang.... forever!
Did anyone at proxmox test this before including it in Proxmox 8?
If so, what arg: line and host command line was used?
I tried the arg: line I used from a pve7.4 VM (which worked fine once virtiofsd was running in 7.x) but it does not appear to work with version 8.
I see the example here, but once I put the qemu-system arguments in the args: line, virtiofsd starts in my hookscript, then dies. If I add --debug-level trace, no additional or relevant data is added that helps troubleshoot. It just dies. I've tried so many variations I've lost track... none of them have worked.
I currently have the following in my VMID.conf file:
Code:
args: -chardev socket,id=char0,path=/run/praxis-rcc.sock -device vhost-user-fs-pci,queue-size=1024,chardev=char0,tag=rcc -object memory-backend-file,id=mem,size=32G,mem-path=/dev/shm,share=on -numa node,memdev=memAnd run the following from my hookscript in the pre-start phase of the hookscript:
Code:
nohup /usr/libexec/virtiofsd --log-level trace --socket-path=/run/praxis-rcc.sock --shared-dir=/mnt/home/rcc --announce-submounts --inode-file-handles=mandatory
&> /var/tmp/virtiofsd-start.out &All I get in my virtiofs-start.out file is:
Code:
# cat /var/tmp/virtiofsd-start.out
nohup: ignoring input
[2023-07-06T00:29:41Z DEBUG virtiofsd::passthrough::mount_fd] Creating MountFd: mount_id=524, mount_fd=10
[2023-07-06T00:29:41Z DEBUG virtiofsd::passthrough::mount_fd] Dropping MountFd: mount_id=524, mount_fd=10
[2023-07-06T00:29:41Z INFO virtiofsd] Waiting for vhost-user socket connection...Before it dies on qmstart with no further output.
If I restart virtiofsd by hand, which appears to stay running using the same command line but started manually, then try to mount a directory in a guest (Ubuntu 23.04) with:
Code:
$ sudo mount -t virtiofs rcc /mnt/test/rcc
$ echo $?
0But then:
ls /mnt/test/rcc
will hang.... forever!
Did anyone at proxmox test this before including it in Proxmox 8?
If so, what arg: line and host command line was used?
