[Bug] NBD Backend: pct resize multiplies target size by 1024 (GB becomes TB) #57
Description
Pardon my English for the title, I don't know how to explain it better. If you wanted to resize a container from 8GB to 12GB ( by adding 4GB ), the plugin instead converts the original "8GB" to "8TB" and our intended "+4G" is converted to "+4T". In the end, your container ends up with 8T + 4T = 12TB
Setup:
pve-version:
root@gamohosthp:~# pveversion -v
proxmox-ve: 9.1.0 (running kernel: 6.17.2-1-pve)
pve-manager: 9.1.1 (running version: 9.1.1/42db4a6cf33dac83)Running plugin version pve-moosefs v0.1.11 (beta).
Issue
I'm copy-pasting the creation steps over from #52 , as the issue is related but occurs with the NBD backend
Create a Proxmox LXC with everything default settings. You can do this either through the Web GUI of Proxmox, or use CLI.
For visibility I'll show the command I used with CLI and console logs.
# This creates a Debian 13 LXC, with the rootfs being in a MooseFS mountpoint with size=8GB.
root@gamohosthp:/# pct create 500 /mnt/pve/gamoblock/template/cache/debian-13-standard_13.1-2_amd64.tar.zst --rootfs GamoBlock:8
Plugin "PVE::Storage::Custom::MooseFSPlugin" is implementing an older storage API, an upgrade is recommended
Formatting '/mnt/pve/gamofs/images/500/vm-500-disk-0.raw', fmt=raw size=8589934592 preallocation=off
Creating filesystem with 2097152 4k blocks and 524288 inodes
Filesystem UUID: 3b5fae63-a641-4eb9-9e7d-55e9962e67d7
Superblock backups stored on blocks:
32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632
extracting archive '/mnt/pve/gamofs/template/cache/debian-13-standard_13.1-2_amd64.tar.zst'
Total bytes read: 551505920 (526MiB, 155MiB/s)
Detected container architecture: amd64
Creating SSH host key 'ssh_host_ed25519_key' - this may take some time ...
done: SHA256:NqMWvxhuZoJ+TwAbM1kz16RehL7f5EtV2z+5LV4f2fg root@localhost
Creating SSH host key 'ssh_host_ecdsa_key' - this may take some time ...
done: SHA256:YQThMJ7ZbIWVHwirv0fOR6LE3O1LaS5Grwf5MdGl2Xc root@localhost
Creating SSH host key 'ssh_host_rsa_key' - this may take some time ...
done: SHA256:cduMXNxXIASyNxDiyo4KjSh0RR2ns1z2xBGO5tJ+yjo root@localhostEnd of copy-pasted stuff
Now, to reproduce the bug, execute a pct resize pct resize 500 rootfs +4G. We are increasing a container's rootfs partition from 8GB to 12GB
root@gamohosthp:~# pct resize 500 rootfs +4G
Plugin "PVE::Storage::Custom::MooseFSPlugin" is implementing an older storage API, an upgrade is recommended
stop block device: (/dev/mfs/mfsmaster.gamo_9421__images_500_vm-500-disk-0->/dev/nbd0 : MFS://images/500/vm-500-disk-0 : 8.000GiB)
started block device: (/dev/mfs/mfsmaster.gamo_9421__images_500_vm-500-disk-0->/dev/nbd0 : MFS://images/500/vm-500-disk-0 : 12288.000GiB)
e2fsck 1.47.2 (1-Jan-2025)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/nbd0: 21929/524288 files (0.2% non-contiguous), 202028/2097152 blocks
resize2fs 1.47.2 (1-Jan-2025)
Resizing the filesystem on /dev/nbd0 to 3221225472 (4k) blocks.
The filesystem on /dev/nbd0 is now 3221225472 (4k) blocks long.This line shows
started block device: (/dev/mfs/mfsmaster.gamo_9421__images_500_vm-500-disk-0->/dev/nbd0 : MFS://images/500/vm-500-disk-0 : 12288.000GiB)
the volume is now internally 12TB long lol. Although the LXC configuration still shows size=12G, the internal volume size is incorrect. If I later move this volume using pct move-volume, the container config updates to reflect the real size (size=12T), allowing the container to use that massive amount of space lol.
pct move-volume 500 rootfs GamoFS: ("GamoFS" is my MooseFS FUSE mountpoint)
The issue seems related to this line, but I did not test further.
Line 1044 in c84ee99