Unassigned Devices - Managing Disk Drives and Remote Shares Outside of The Unraid Array

[JorgeB]

22 minutes ago, dlandon said:

Pool disks would have the same mount point?

Yep.

[dlandon]

1 minute ago, JorgeB said:

Yep.

Is there a way for UD to detect that disks are in a pool?

[JorgeB]

3 minutes ago, dlandon said:

Is there a way for UD to detect that disks are in a pool?

Don't know if you can use it but at boot time all btrfs filesystems are scanned, e.g.:
 

Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 422c457c-5e12-4be7-8422-a661e00c7ffc devid 2 transid 5013 /dev/sdi1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid ae12a64a-ab83-429e-a449-e02b0d1a3453 devid 2 transid 4813 /dev/sdf1 scanned by udevd (1512)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 99c55241-a0cf-4003-b66d-73512686ce4e devid 1 transid 34 /dev/sdc1 scanned by udevd (1679)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 422c457c-5e12-4be7-8422-a661e00c7ffc devid 1 transid 5013 /dev/sdb1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid ae12a64a-ab83-429e-a449-e02b0d1a3453 devid 1 transid 4813 /dev/sdg1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 56d7a584-5bfd-4d97-94bf-9a0ddc798c65 devid 1 transid 13504 /dev/sde1 scanned by udevd (1679)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 99c55241-a0cf-4003-b66d-73512686ce4e devid 2 transid 34 /dev/sdh1 scanned by udevd (1512)

 

Devices with the same fsid are from the same pool.

[Vr2Io]

7 minutes ago, JorgeB said:

Don't know if you can use it but at boot time all btrfs filesystems are scanned, e.g.:
 

Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 422c457c-5e12-4be7-8422-a661e00c7ffc devid 2 transid 5013 /dev/sdi1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid ae12a64a-ab83-429e-a449-e02b0d1a3453 devid 2 transid 4813 /dev/sdf1 scanned by udevd (1512)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 99c55241-a0cf-4003-b66d-73512686ce4e devid 1 transid 34 /dev/sdc1 scanned by udevd (1679)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 422c457c-5e12-4be7-8422-a661e00c7ffc devid 1 transid 5013 /dev/sdb1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid ae12a64a-ab83-429e-a449-e02b0d1a3453 devid 1 transid 4813 /dev/sdg1 scanned by udevd (1511)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 56d7a584-5bfd-4d97-94bf-9a0ddc798c65 devid 1 transid 13504 /dev/sde1 scanned by udevd (1679)
Dec 7 15:37:51 Tower11 kernel: BTRFS: device fsid 99c55241-a0cf-4003-b66d-73512686ce4e devid 2 transid 34 /dev/sdh1 scanned by udevd (1512)

 

Devices with the same fsid are from the same pool.

 or use "btrfs fi show"

 

[TheJavarian]

2 hours ago, dlandon said:

Please do the following:

• Update to the latest UD (2021.12.08).
• Try to mount the device again.
• Post diagnostics and the contents of the file /var/state/unassigned.devices/share_names.json.

I've added some information to the log message about the duplicate share name to help find this issue.

Thank you for helping me with this issue.

 

I updated UD to 2021.12.08. Still it does not mount device.

Here are the e files.

milkyway-diagnostics-20211209-0943.zip share_names.json

[dlandon]

40 minutes ago, TheJavarian said:

Thank you for helping me with this issue.

 

I updated UD to 2021.12.08. Still it does not mount device.

Here are the e files.

milkyway-diagnostics-20211209-0943.zip 227.38 kB · 1 download share_names.json 131 B · 1 download

Update to the latest UD release 2021.12.09 and see if the issues is solved.  If not, post diagnostics and share_names.json once more.

[dlandon]

For those of you with pools in UD, go to the UD Settings and set "Check Unassigned Devices Share Names?" to 'No'.  UD will not check for duplicate UD share names.

[dlandon]

3 hours ago, stev067 said:

 

I also have been experiencing this refusal to mount, since yesterday. I didn't change anything, and I've always been able to just mount/unmount this drive whenever I wanted. I found that stopping the array was enough to get it to mount, rather than rebooting.

This implies that you have a UD device mount point (share name) the same as one in Unraid.  This is not a good idea and will potentially cause a conflict if you share the UD device.  You need to change the mount point on the UD device.

[TheJavarian]

55 minutes ago, dlandon said:

Update to the latest UD release 2021.12.09 and see if the issues is solved.  If not, post diagnostics and share_names.json once more.

Thank you very much,

 

It worked!. I could mount the device without any further problem.

 

[Flick]

At first blush I thought the new "mark as 'Array'" feature was great, as I can see the benefits. However, I have a unique use case and it's tripping me up.

 

Basically, I have an old array drive that was pulled out of the array and replaced with a larger drive. I have a need (don't ask, seriously) to mount this old drive as Unassigned and copy some files from it. I've done this a couple of times in the past, no big deal. Now, however, the new feature blocks me from mounting it. I've poked around for an override option but I don't see one.

 

Suggestions?

 

Thank you, as always, for all your hard work on this! We all appreciate you!

[dlandon]

3 minutes ago, Flick said:

At first blush I thought the new "mark as 'Array'" feature was great, as I can see the benefits. However, I have a unique use case and it's tripping me up.

 

Basically, I have an old array drive that was pulled out of the array and replaced with a larger drive. I have a need (don't ask, seriously) to mount this old drive as Unassigned and copy some files from it. I've done this a couple of times in the past, no big deal. Now, however, the new feature blocks me from mounting it. I've poked around for an override option but I don't see one.

 

Suggestions?

 

Thank you, as always, for all your hard work on this! We all appreciate you!

This shouldn't show up unless Unraid doesn't see the disk as unassigned which means it's assigned to the array or pool.  It should not be happening to you.

 

Give me a screen shot of the array disks and UD disks.

[Flick]

56 minutes ago, dlandon said:

This shouldn't show up unless Unraid doesn't see the disk as unassigned which means it's assigned to the array or pool.  It should not be happening to you.

 

Give me a screen shot of the array disks and UD disks.

So after posting I discovered the "Change UUID" function. It didn't seem to work, but when I went back in just now to get a screenshot for you, the drive now shows as Mountable.

 

I'll need to do this to one more drive, so I'll capture a screenshot there to help you troubleshoot more, if you like, but I'm now content that there is indeed a workaround.

 

Thanks for the super fast follow-up!

[dlandon]

52 minutes ago, Flick said:

So after posting I discovered the "Change UUID" function. It didn't seem to work, but when I went back in just now to get a screenshot for you, the drive now shows as Mountable.

 

I'll need to do this to one more drive, so I'll capture a screenshot there to help you troubleshoot more, if you like, but I'm now content that there is indeed a workaround.

 

Thanks for the super fast follow-up!

Why do you say the change UUID didn't work?  Look in the log and it will tell you if it was successful.

[Flick]

On 12/10/2021 at 11:51 AM, dlandon said:

Why do you say the change UUID didn't work?  Look in the log and it will tell you if it was successful.

I believe you missed that I said "seemed not to work." It took a few minutes for the system to recognize the new UUID and then the "Array" label went away and I was able to mount.

[cmon_google_wtf]

Forgive me, I tried search but came up not quite understanding.

Use case: Simple, full disk pass-through to an Ubuntu VM. Performance is critical, so if there's a better option than UD, I'm all ears.

I originally installed the OS using /dev/disk/by-label, with UD marking the device as passed through. After a reboot of the VM, by-label/X was gone, so I remounted in the VM as by-id, passing the first partition (/dev/sdX1) instead. From searching the thread, I found that this (sdX) might not be the best solution, and to use a devX mount instead, but I cannot for the life of me figure out how to both find and pass that devX drive. (dev 1, 2, etc I can see in the UI - how do I "mount" and pass that to the VM?)

 

Thoughts?

[dlandon]

3 hours ago, omninewb said:

Forgive me, I tried search but came up not quite understanding.

Use case: Simple, full disk pass-through to an Ubuntu VM. Performance is critical, so if there's a better option than UD, I'm all ears.

I originally installed the OS using /dev/disk/by-label, with UD marking the device as passed through. After a reboot of the VM, by-label/X was gone, so I remounted in the VM as by-id, passing the first partition (/dev/sdX1) instead. From searching the thread, I found that this (sdX) might not be the best solution, and to use a devX mount instead, but I cannot for the life of me figure out how to both find and pass that devX drive. (dev 1, 2, etc I can see in the UI - how do I "mount" and pass that to the VM?)

 

Thoughts?

I'm not clear on what you are trying to do.

 

When a disk is physically passed through to a VM, you don't want UD to mount it so you mark it as passed through so UD won't touch the disk.  The VM will then mount the physical disk and use it.

 

If you are trying to have UD mount the disk and have the VM use it, you would have UD auto mount it and share it with samba.  Then use /mnt/disks/mountpoint to access it, or access it at //tower/mountpoint depending on the VMs use case.

 

You don't want UD and your VM mounting the physical disk at the same time.  You will corrupt the file system.

[cmon_google_wtf]

6 hours ago, dlandon said:

I'm not clear on what you are trying to do.

 

When a disk is physically passed through to a VM, you don't want UD to mount it so you mark it as passed through so UD won't touch the disk.  The VM will then mount the physical disk and use it.

 

If you are trying to have UD mount the disk and have the VM use it, you would have UD auto mount it and share it with samba.  Then use /mnt/disks/mountpoint to access it, or access it at //tower/mountpoint depending on the VMs use case.

 

You don't want UD and your VM mounting the physical disk at the same time.  You will corrupt the file system.

Understood, and a mis-communication on where I am. The only time I have had UD mount it, was in the initial setup of the disk, when I set the `by-label` and mounted it in the VM using that path.

After the OS install completed and I rebooted the VM, UD is no longer mounting it (passthrough option checked), the and the `by-label` for that disk is gone from `/dev/disks/by-label`, but unless I mount (in the VM options) either the first partition or the /dev/sdX option, I cannot get past the bootloader. 

 

This prompted the question as to if I was missing where I could passthrough and/or mount the /devX.

[dlandon]

1 hour ago, omninewb said:

This prompted the question as to if I was missing where I could passthrough and/or mount the /devX.

The devX designation is assigned by Unraid to the UD device similar to naming a pool device.  You can't mount it directly.

[mgeem]

Does anyone know why I'd see that status for a previously functioning device? 

[dlandon]

UD thinks the disk is an array disk that has disconnected and reconnected with a different device designation.  What is the history of this device?  Always been a UD device?

 

Post your diagnostics.

[mgeem]

it was working perfectly fine earlier today, it has always been a UD device

unraid-diagnostics-20211213-1418.zip

[dlandon]

15 minutes ago, mgeem said:

it was working perfectly fine earlier today, it has always been a UD device

unraid-diagnostics-20211213-1418.zip 482.33 kB · 0 downloads

Looks to be a bug in UD.  I'll issue an update tomorrow to fix it.

[mgeem]

Appreciate your swift response - thank you for your help.

[R__]

Hi, I'm not quite sure how I got in this situation, but two of my disks have the same mount point, which leads to obvious issues when mounting/unmounting (they both mount/unmount at the same time, but only one of the disk's contents shows up under /mnt/disks/mountpointname.

 

Is there a way to fix this? These are both disks in an external USB enclosure. The other 2 disks in the same enclosure both have unique mount points.

 

Thanks!

[dlandon]

1 hour ago, R__ said:

Hi, I'm not quite sure how I got in this situation, but two of my disks have the same mount point, which leads to obvious issues when mounting/unmounting (they both mount/unmount at the same time, but only one of the disk's contents shows up under /mnt/disks/mountpointname.

 

Is there a way to fix this? These are both disks in an external USB enclosure. The other 2 disks in the same enclosure both have unique mount points.

 

Thanks!

Post a UD screen shot and your diagnostics.