Convert VM on ZFS to Vmware

[Fafa24]

Hi,

This is my first post on this board. Please be nice to me.

I have a single Proxmox server in my homelab with zfs. I want to convert a VM to VMware as a test.

For ZFS I'm not sure which VM disk I need to convert.

find / | grep vm-101-disk-0 - outputs

root@proxmox1:/dev/zvol/datastore# find / | grep vm-101-disk-0
/dev/zvol/datastore/vm-101-disk-0-part3
/dev/zvol/datastore/vm-101-disk-0-part1
/dev/zvol/datastore/vm-101-disk-0-part2
/dev/zvol/datastore/vm-101-disk-0
/dev/datastore/vm-101-disk-0-part3
/dev/datastore/vm-101-disk-0-part1
/dev/datastore/vm-101-disk-0-part2
/dev/datastore/vm-101-disk-0
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part1
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part1/b230:1
/run/udev/links/datastore\x2fvm-101-disk-0-part1
/run/udev/links/datastore\x2fvm-101-disk-0-part1/b230:1
/run/udev/links/datastore\x2fvm-101-disk-0-part2
/run/udev/links/datastore\x2fvm-101-disk-0-part2/b230:2
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part2
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part2/b230:2
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part3
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0-part3/b230:3
/run/udev/links/datastore\x2fvm-101-disk-0-part3
/run/udev/links/datastore\x2fvm-101-disk-0-part3/b230:3
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0
/run/udev/links/zvol\x2fdatastore\x2fvm-101-disk-0/b230:0
/run/udev/links/datastore\x2fvm-101-disk-0
/run/udev/links/datastore\x2fvm-101-disk-0/b230:0
root@proxmox1:/dev/zvol/datastore#

Would I need to convert the disk in bold? I understand in ZFS it is not a disk file.

qemu-img convert -f raw r/dev/zvol/datastore/vm-101-disk-0 -O vmdk zappix-neu.vmdk

If everything is correct, will the qemu-img command convert the part files too?

Thanks a lot.

 

[LnxBil]

Would I need to convert the disk in bold?

Yes.

I understand in ZFS it is not a disk file.

Technically, it's a block device (which is also a file, because everything is a file in Linux).

qemu-img convert -f raw r/dev/zvol/datastore/vm-101-disk-0 -O vmdk zappix-neu.vmdk

If everything is correct, will the qemu-img command convert the part files too?

Yes, that looks ok.

 

[Fafa24]

Thank you for your reply. Appreciate it

 

[Deleted member 205422]

Technically, it's a block device (which is also a file, because everything is a file in Linux).

I would be careful about these assumptions with ZFS, not everything that is a file is block device.

# zfs list -t filesystem -o name,used
NAME                      USED
rpool                    46.7G
rpool/ROOT               2.02G
rpool/ROOT/pve-1         2.02G
rpool/data                 96K
rpool/subvol-101-disk-0   936M
rpool/subvol-102-disk-0  6.81G
rpool/subvol-105-disk-0  9.04M
rpool/subvol-106-disk-0   443M

# zfs list -t volume -o name,used
NAME         USED
rpool/swap  34.0G
root@pve5:~#

There's datasets that are filesystem type and there's zvols which are actually block devices. If it was block device in the regular sense of the word a swap could be created on it. PVE does not create ZVOLs. ZVOLs are linked from the pool's subpath in /dev and mixed inbetween other datasets.

EDIT: Mistakenly mixed up statement on VMs and CTs in PVE.

For the more curious: https://www.youtube.com/watch?v=BIxVSqUELNc

But take my post as a footnote only.

 

[LnxBil]

I would be careful about these assumptions with ZFS, not everything that is a file is block device.

It's not an assumption. It's a fact that the OP's bold file is a block device. Datasets are not represented in /dev/zvol, only zvols ... hence the name.

 

[Deleted member 205422]

It's not an assumption. It's a fact that the OP's bold file is a block device. Datasets are not represented in /dev/zvol, only zvols ... hence the name.

This statement is correct, however I found the terse reaction on the "file" aspect of it a bit quirky. The OP, I believed, wanted to confirm it's not a VM image file (in the casual sense of the word) he was dealing with.

It is also true that it was a block device he was getting converted because everything in /dev/zvol is.

I also agree that everything in Linux is represented by a file descriptor.

I am not sure statements with broadly used terms like "file" or "ZFS" are self-explanatory. I do not know if OP was interested, but others might find this post helpful later on (including with our nitpicking).

EDIT: Went back and checked what caused the reaction, fixed the wrong statement, thanks for reaction.

 

[Fafa24]

qemu-img convert -f raw r/dev/zvol/datastore/vm-101-disk-0 -O vmdk zappix-neu.vmdk

To clarify - Will the qemu-img convert command recognize there are Part1, Part2 and Part3?

 

[LnxBil]

To clarify - Will the qemu-img convert command recognize there are Part1, Part2 and Part3?

Yes, those are the partitons inside of the disk image. The resulting vmdk will include all 3 parts AND the partition table.

 

[Fafa24]

Thanks for confirming @LnxBil - I assumed so, but was not sure

 

[Deleted member 205422]

Thanks for confirming @LnxBil - I assumed so, but was not sure

In case you wondered why that is:

Spoiler

# zfs list -o name,type

NAME                     TYPE
rpool                    filesystem
rpool/ROOT               filesystem
...
rpool/swap               volume
rpool/vm-201-disk-0      volume

# ls -la /dev/rpool/vm-201-disk*

lrwxrwxrwx 1 root root 7 Nov 11 17:01 /dev/rpool/vm-201-disk-0 -> ../zd16
lrwxrwxrwx 1 root root 9 Nov 11 17:01 /dev/rpool/vm-201-disk-0-part1 -> ../zd16p1
lrwxrwxrwx 1 root root 9 Nov 11 17:01 /dev/rpool/vm-201-disk-0-part2 -> ../zd16p2
lrwxrwxrwx 1 root root 9 Nov 11 17:01 /dev/rpool/vm-201-disk-0-part5 -> ../zd16p5

# ls -la /dev/zvol/rpool/vm-201-disk*

lrwxrwxrwx 1 root root 10 Nov 11 17:01 /dev/zvol/rpool/vm-201-disk-0 -> ../../zd16
lrwxrwxrwx 1 root root 12 Nov 11 17:01 /dev/zvol/rpool/vm-201-disk-0-part1 -> ../../zd16p1
lrwxrwxrwx 1 root root 12 Nov 11 17:01 /dev/zvol/rpool/vm-201-disk-0-part2 -> ../../zd16p2
lrwxrwxrwx 1 root root 12 Nov 11 17:01 /dev/zvol/rpool/vm-201-disk-0-part5 -> ../../zd16p5

# ls -la /dev/zd16*

brw-rw---- 1 root disk 230, 16 Nov 11 17:01 /dev/zd16
brw-rw---- 1 root disk 230, 17 Nov 11 17:01 /dev/zd16p1
brw-rw---- 1 root disk 230, 18 Nov 11 17:01 /dev/zd16p2
brw-rw---- 1 root disk 230, 21 Nov 11 17:01 /dev/zd16p5

# ls -la /sys/devices/virtual/block/zd16/

total 0
drwxr-xr-x 11 root root    0 Nov 11 16:45 .
drwxr-xr-x 12 root root    0 Nov  5 02:52 ..
-r--r--r--  1 root root 4096 Nov 11 17:15 alignment_offset
lrwxrwxrwx  1 root root    0 Nov 11 17:15 bdi -> ../../bdi/230:16
-r--r--r--  1 root root 4096 Nov 11 17:15 capability
-r--r--r--  1 root root 4096 Nov 11 17:01 dev
-r--r--r--  1 root root 4096 Nov 11 17:15 discard_alignment
-r--r--r--  1 root root 4096 Nov 11 17:15 diskseq
-r--r--r--  1 root root 4096 Nov 11 17:15 events
-r--r--r--  1 root root 4096 Nov 11 17:15 events_async
-rw-r--r--  1 root root 4096 Nov 11 17:15 events_poll_msecs
-r--r--r--  1 root root 4096 Nov 11 17:15 ext_range
-r--r--r--  1 root root 4096 Nov 11 17:15 hidden
drwxr-xr-x  2 root root    0 Nov 11 17:01 holders
-r--r--r--  1 root root 4096 Nov 11 17:15 inflight
drwxr-xr-x  2 root root    0 Nov 11 17:01 integrity
drwxr-xr-x  2 root root    0 Nov 11 17:01 power
drwxr-xr-x  2 root root    0 Nov 11 16:45 queue
-r--r--r--  1 root root 4096 Nov 11 17:15 range
-r--r--r--  1 root root 4096 Nov 11 17:15 removable
-r--r--r--  1 root root 4096 Nov 11 17:15 ro
-r--r--r--  1 root root 4096 Nov 11 16:45 size
drwxr-xr-x  2 root root    0 Nov 11 17:01 slaves
-r--r--r--  1 root root 4096 Nov 11 17:15 stat
lrwxrwxrwx  1 root root    0 Nov 11 16:45 subsystem -> ../../../../class/block
drwxr-xr-x  2 root root    0 Nov 11 17:01 trace
-rw-r--r--  1 root root 4096 Nov 11 16:45 uevent
drwxr-xr-x  5 root root    0 Nov 11 17:01 zd16p1
drwxr-xr-x  5 root root    0 Nov 11 17:01 zd16p2
drwxr-xr-x  5 root root    0 Nov 11 17:01 zd16p5

# cat /lib/udev/rules.d/60-zvol.rules

# Persistent links for zvol
#
# persistent disk links: /dev/zvol/dataset_name
# also creates compatibility symlink of /dev/dataset_name

KERNEL=="zd*", SUBSYSTEM=="block", ACTION=="add|change", PROGRAM=="/lib/udev/zvol_id $devnode", SYMLINK+="zvol/%c %c"

# head /lib/udev/rules.d/60-persistent-storage.rules

# do not edit this file, it will be overwritten on update

# persistent storage links: /dev/disk/{by-id,by-uuid,by-label,by-path}
# scheme based on "Linux persistent device names", 2004, Hannes Reinecke <hare@suse.de>

ACTION=="remove", GOTO="persistent_storage_end"
ENV{UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG}=="1", GOTO="persistent_storage_end"

SUBSYSTEM!="block|ubi", GOTO="persistent_storage_end"
KERNEL!="loop*|mmcblk*[0-9]|msblk*[0-9]|mspblk*[0-9]|nvme*|sd*|sr*|vd*|xvd*|bcache*|cciss*|dasd*|ubd*|ubi*|scm*|pmem*|nbd*|zd*", GOTO="persistent_storage_end"