vm disk parts

[Aleksej]

Hello.

Can somebody explain.
I have ZFS pool storage. under /dev/zvol/mypool i saw vm-10115-disk-0

after i disabled atime

zfs set atime=off mypool

under /dev/zvol/mypool i see

lrwxrwxrwx 1 root root 11 Jun  9 19:34 vm-10115-disk-0 -> ../../zd112
lrwxrwxrwx 1 root root 13 Jun  9 19:34 vm-10115-disk-0-part1 -> ../../zd112p1
lrwxrwxrwx 1 root root 13 Jun  9 19:34 vm-10115-disk-0-part2 -> ../../zd112p2
lrwxrwxrwx 1 root root 13 Jun  9 19:34 vm-10115-disk-0-part3 -> ../../zd112p3

vm-10115-disk-0 vm-10115-disk-0-part1 vm-10115-disk-0-part2...

before i used simple dd if=/dev/zvol/mypool/vm-10115-disk-0 | nc 1.1.1.1 1111 to "migrate" vm. but what part i need to use now?

 

[Kingneutron]

dd and netcat aren't how you (correctly) migrate a VM... Back it up using Proxmox integrated backup functionality, and restore to where you want it - or cluster and setup HA

 

[Aleksej]

Actually i asked about what is partX in zfs pool.

Why not use dd via netcat or ssh?
Offtop:
Or maybe you can suggest better way to move whole VM without having pbs?

 

[Kingneutron]

You don't need pbs to backup and restore a VM, as I said in previous post just use integrated backup to separate media or NAS

 

[Aleksej]

In my case it is a bit expensive.
I have nodes in one datacenter and yes i have backup server but in different place.
Link between nodes is 1Gbps and 10Gbps, but link to backup space is only 100Mbps.
Actually migrating VMs are not often, i try to divide them by resources to proper nodes.
But sometimes it happens.

Before ZFS i used LVM Thin and migrate was very simple.
dd if=mytv bs=2M | ssh.... dd of=mynewthinvol bs=2M and after that i could delete vm on "old" host.

At first with zfs was the same. All worked fine. But few time ago a faced that under /dev/zvol/zpool there is "-partX" volumes.

Nowhere in zfs docs are not desribed that vloume can be/will be divided into parts.
And even in zfs docs written that volume can be accessed as block device under /dev/zvol/zpool/myvolume

 

[Paleo]

In Proxmox UI you can select your VM, then click um Backup (Proxmox VE backup, not Proxmox Backup Server) and click on Backup Now. After that you simply move back up file to the new server, place it in backup up folder, go in new server UI click on Backup and then in Restore.
Your migration is done. I believe this was what Kingneutron was trying to say.

 

[Aleksej]

Yes, i agree. But for this operation to be succesfult VM must be powered off.
As for me, if i have to power off VM, why to do backups if i can simply copy whole block device)
In any case make backup - transfer - restore need much more time.
Even more, if i place zstd in combination with nc and dd it is slower than copy whole disk.
Only once it was faster because on 400Gb disk data was only 10Gb. So zstd was fast.


But lets return to my question. As for me, everyone decides himself how to transfer VM.

But, what is strange ....disk-0,..... disk-0-part1, etc?
It is only on ZFS.

P. S. Even with compression off.