Moved zvol, different size afterwards

[chuck32]

I recently switched to trueNAS as my hypervisor. With some trouble I was able to install Home Assistant and that seems to be working too. Today I took two 2.5" SSDs (with my pci sata card) and created a mirror to store all my apps and VM data. I needed to move the home assistant zvol for that. I didn't want all the I/O on my storage pool, which consists of HDDs.

I followed this guide to install HA (write the raw HassOS image to zvol) and allocated 50 gb to the zvol. To move to the new SSD pool I used:

Code:

zfs snapshot -r pool1/homeassistant@relocate
zfs send pool1/homeassistant@relocate | zfs receive -v pool2/homeassistant

The new zvol has only 4 gb data written instead of 35 gb (also for some reason now 55 gb are allocated, I deleted the @relocate snapshot via the GUI).

Home assistant is working and recognizes the 50 gb allocation:

Can I safely delete the old zvol or is there still some dependency on it?

 

[sretalla]

Provisioning type: Sparse means you won't necessarily be using all the blocks...

You'll see that the larger/original was Provisioning type: Thick (write/own all the blocks even if not used yet)

I think that mostly explains the difference.

 

[chuck32]

So basically I can ignore the difference in size then and delete the original zvol? I'll create a snapshort first anyway to just be sure.

I posted the commands I used, I didn't intentionally change the provisioning type. Nothing wrong with sparse I guess?

I didn't really catch that detail, thank you! Now that you pointed it out it's blatant though.

 

[sretalla]

It's highly unlikely that zfs send | recv only "partially worked"...

If it works at all, it's going to be a full and complete "copy" of what was on the original.

I would say safe to do what you want with the old one if you're running with the new one.

I also wanted to say that compression probably accounts for the other large difference... thick provisioned would be lots of blocks of all zero, so compresses well.

 

[shoulders]

I know this is an old thread.

When you copy the ZVol by using a snapshot, ZFS copies it as 'Thin' provisioned probably to save bandwidth. There might be a switch to change this, but I used the following to fix mine:

Correct Target ZVol disk usage

Code:

sudo zfs set refreservation=auto MyPoolA/Virtual_Disks/MyDatasetA

Altering the other related ZVol size settings did not have the desired effect if you are wondering, I only had partial success with them.