SOLVED Pool Creation Stuck at "Fetching Data..." (Fails with old MDRAID drives)

[jrwrigh]

Hello, I'm a brand new TrueNAS user trying to setup my first server. I've installed SCALE and booted the system. I'm now trying to create a pool.

I'm trying to create a mirror pool over my 3 2TB drives. I've gone through the steps specified here. However, it's been stuck on "Fetching data..." for about 2 hours now.

I remember with my previous NAS that the RAID initialization took several hours, but I had heard that ZFS should only take a few minutes (including some YouTube videos showing it taking that long for SCALE). Additionally, the lack of progress information is a bit disconcerting. The second I click off of the "Create Pool" window, it just exits and it doesn't appear that anything is happening in the background and I can request creating another pool with the exact same drives.

So I guess my questions are:
- How long should the pool creation take (ballpark anyways)?
- Is there a way to "background" the pool creation so that I can do other things (explore the interface, install apps, etc.)?

System Info:
Motherboard: ASUS
CPU: Intel Core i5-4590
RAM: 16GB DDR3
Hard Drives: 1x 2TB WD Red WD20EFRX, 2x 2TB Seagate Barracuda ST2000DM008

 

[jrwrigh]

Oh, and the OS version is: TrueNAS-SCALE-22.02.3

 

[jrwrigh]

In trying to create the pool manually in the shell (using zpool create mirror sda sdb sdc),

Code:

/dev/sda is in use and contains and unknown filesystem.
/dev/sdb is in use and contains and unknown filesystem.
/dev/sdc is in use and contains and unknown filesystem.

The drives were previously used in a RAID setup (but I want to wipe that away). Not sure if that has any impact on the setup process.

 

[jrwrigh]

This Stack Exchange answer mentions needing to let deactivate mdraid in order to get the drives working correctly. Maybe that's the issue? And if so, should I open a GH issue to make the GUI interface recognize this and error out the process (assuming the GUI is running into this issue and then does nothing afterwards)?

 

[NugentS]

wipe the disks before putting them in the NAS. Windows diskpart & clean would be one option. Just make sure you get the right disk

 

[jrwrigh]

I daily-drive Linux, so I got it to work via the shell in the TrueNAS GUI (System Settings>Shell) via:

Code:

wipefs -af /dev/sd{a,b,c}

mdadm -S /dev/md12{2..7}

The former should remove any of the raid information from the disk, the latter stops the raid service.

For anyone who might find this and try to do the same, note the /dev/sd{a,b,c} and /dev/md12{2..7} will need to be adjusted to whatever the setup of your system is. Also, that /dev/sd{a,b,c} is just shorthand for /dev/sda /dev/sdb /dev/sdc, similar for /dev/md12{2..7}. You can just type each one individually if you feel more comfortable with that.

 

[jrwrigh]

Few extra notes I forgot:

- To determine the names of the drives and the /dev/md* partitions, lsblk is a useful command.

- Also, addressing my "How long should it take?", once I fixed the drives, the pool setup time took less than a minute and it continued working after I had exited out of the window. The "Tasks" button in the top right showed the progress.

- Lastly, I've created a ticket in Jira for this: https://ixsystems.atlassian.net/browse/NAS-118244

 

[anon_y_mous]

Code:

wipefs -af /dev/sd{a,b,c}

mdadm -S /dev/md12{2..7}

For some reason these commands did not work for me. No errors, just did not have the desired effect. However the steps in this article did work. Though I had to manually enter the seek value as the calculation didn't interpret correctly, maybe due to the difference in shell from the article and TrueNAS, not sure. https://support.siliconmechanics.com/portal/en/kb/articles/clearing-raid-metadata

So my steps were:

Code:

dd if=/dev/zero of=/dev/sd{a,b,c} bs=512 count=1024
blockdev --getsz /dev/sd{a,b,c} # take output from this and subtract 1024 manually
dd if=/dev/zero of=/dev/sd{a,b,c} bs=512 seek=<value from previous step> count=1024