freenas 9.1 zfs volume of different drive size

[JunkyMonkey]

Did FreeNAS 9.1 remove the ability to create a zfs volume with different drive sizes? I've been trying to figure out how to create a raidz2 with my 6 drives of different sizes (3x3tb, 2x1tb and 1x2tb) and just wasn't able to whereas I was able to do it in freenas 8.3.1. The slider feature in 9.1 creates a raidz1 with the 3x3tb and a mirror with the 2x1tb while leaving the 2tb as a spare. Please help.

 

[dlavigne]

Yes, the new graphical wizard is designed to prevent users from unknowingly creating un-optimal configurations. A determined user could still make such a configuration from the command line.

 

[Lukas]

Hi there,

I am such a "determined user" and just tried creating a new pool using the command

Code:

zpool create -f data raidz1 ada0 ada1 ada2 ada3

but as a result only get an error message

Code:

cannot create 'test': no such pool or dataset

Any help?

 

[Lukas]

but as a result only get an error message

Code:

cannot create 'test': no such pool or dataset

I solved this by re-adding every disk in a single zpool and after that removing it and checking the "remove data and mark as new" checkbox. After that zpool creation worked just fine.

 

[BillyBob2]

This issue is realy annoying. I have 5 x 2TB disks,and 1 x 3TB disk that I would like to make into a ZFS pool, but not able to make it happen thus far. Why didn't they add a check box, to bypass the restriction?
Regretfully, all of the above suggestions have not worked as of yet. Except for using a 3rd party disk formatting utility, that I will try now and report back on the result.

 

[paleoN]

Except for using a 3rd party disk formatting utility, that I will try now and report back on the result.

No need for 3rd party. It can all be done from the CLI. If the disks have existing metadata on them FreeBSD may be protecting it and you need to clear said metadata first.

 

[BillyBob2]

The issue was that after clearing out all of the disks I kept then getting this message:
"cannot mount '/Tank': failed to create mountpoint"
I then found a post that said the solution to this error was
"

You must auto import using the webgui....​

(Make sure the pool is exported... "zpool export Tank")​

http://forums.freenas.org/threads/freenas-8-2-replacing-a-dead-drive-in-zfs.11202/

 

[paleoN]

"cannot mount '/Tank': failed to create mountpoint"

Oh, so you had a different problem. You need to have the pool mounted under /mnt.

You must auto import using the webgui....​

(Make sure the pool is exported... "zpool export Tank")​

Yes, create the pool, export the pool and import via the GUI.

 

[BillyBob2]

Thank you for the feedback paleoN,

one last question.

How does the pool get mounted in the wrong (non /mnt) location?
Would there be a different way to format the drives, that would make future volume builds with different drives sizes simpler? Not that all of the steps listed above are so complicated once you know what you are doing.
its just you think it would be a little more strait forward.

 

[paleoN]

How does the pool get mounted in the wrong (non /mnt) location?

FreeNAS is an appliance based, embedded OS and not full blown FreeBSD. There's nothing wrong with a non /mnt location except it doesn't work by default under FreeNAS.

Would there be a different way to format the drives, that would make future volume builds with different drives sizes simpler?

Not sure what you are asking here. Unless, you mean from the GUI? No, as the GUI uses whole devices. I can't recall if there was talk of adding a force option back in for different sized drives.

 

[rcstevensonaz]

This issue is realy annoying. I have 5 x 2TB disks,and 1 x 3TB disk that I would like to make into a ZFS pool, but not able to make it happen thus far. Why didn't they add a check box, to bypass the restriction?
Regretfully, all of the above suggestions have not worked as of yet. Except for using a 3rd party disk formatting utility, that I will try now and report back on the result.

I also agree this is annoying. A mixture of 500GB and 640GB drives is not sub-optimum. And, sometimes we are intentionally adding larger drives to a vdev with the expectation of rolling replacement over time. I agree a check-box that allows us to combine different drive sizes would be very helpful.

Also, it would be nice if the system did not work so hard to force its view of optimum. For example, allow an ability to select which disks to use for the vdev. In my case, I am trying to spread across SAS backplanes so that a single failure does not kill a pool. But the only easy way to do it is create the disks using the order FreeNAS wants; then manually pull the drives and insert into the desired configuration layout.

 

[cyberjock]

rcstevensonaz:

Feel free to go tell the developers what you think about this "bug". https://bugs.freenas.org/issues/3274

I've tried to give them info.. but thus far they haven't really listened.

 

[rcstevensonaz]

Thanks Cyberjock. I've updated the bug with my comments; though I am not sure if the answer is to bring back the old ZFS Volume Manager or to fix the new one to support typical use cases. Overall, I am surprised that FreeNAS works against what are typically considered best practices for ZFS: using drives from different vendors in a vdev pool (i.e. the disk sizes might not be identical) and selecting disks that are spread across multiple controllers / SAS expanders, etc.

The reason I wanted to come to FreeNAS was the strength of the community, the support for GPT labels in ZFS vdevs and the ongoing development. But instead, I was surprised to learn that I cannot mange the process of building intentional vdevs unless I revert to a much older version.

 

[cyberjock]

Crazy huh? I think it's stupid too... LOL

 

[tzd]

Hi I'm trying to understand what this means in my setup. I currently have 3x 2TB disks in RAID-Z1, about 4TB of usable storage.

Does this mean that in 9.1, if one of my disks fail in the future, my only option is to replace it with a 2TB disk? If I try to replace it with a 3TB disk it would fail?



Sent from my iPad using Tapatalk

 

[Dusan]

Does this mean that in 9.1, if one of my disks fail in the future, my only option is to replace it with a 2TB disk? If I try to replace it with a 3TB disk it would fail?

No, disk replacements are unaffected. The restriction only applies when you create a new pool.

 

[tzd]

No, disk replacements are unaffected. The restriction only applies when you create a new pool.

Great thanks! Now I can upgrade with peace of mind.

 

[tzd]

rcstevensonaz:

Feel free to go tell the developers what you think about this "bug". https://bugs.freenas.org/issues/3274

I've tried to give them info.. but thus far they haven't really listened.

I attended a FreeNAS presentation last night and the iX developers confirmed that the old volume manager will be back as an option in the next release 9.2.1. A beta might be out soon.

Thanks cyberjock for your persistence in pushing this bug on them! This is a functionality that a lot of us need.



Sent from my iPad using Tapatalk