In trial - Can not get a 250GB SSD as cache to format no matter what.

[IceWater]

Any assistance is appreciated. 

I am evaluating Unraid as a potential NAS / VM & Docker host solution for my home lab. 
I have a few drives in (just as a trial concept) and my 250 GB WD Blue SSD which is set as a cache device will not format. 
All I keep seeing is "Unmountable: not mounted" 

Things I have tried:

• Format with "File system type" set to Auto, xfs, btrfs
• Stopping the array, stating and trying again
• Different port on mobo & new sata cable
• Tried blkdiscard blkdiscard: cannot open /dev/sdb: Device or resource busy

 

Drive is verified functioning when I plug it into an external dock to my windows box. 

Any assistance is appreciated. 


Update: I was able to resolve this issue after pulling the drive and running it through Disk Part with the "Clean" command using an external docking station with my windows box. 

Steps:
1.) Install drive into computer or use external dock to mount

2.) Hit Window + R and in the run box type diskpart than hit ctrl+shit+enter
3.) In the new terminal windows type List Disk

4.) Select your disk typing out the command  select disk X where X is your number of disk, be careful not to select your operating system or other disks. 

5.) Type clean
6.) Type exit.
7.) Remove the drive and re-install in Unraid box.

I hope this helps someone else.  

 

 

Edited April 25, 2021 by IceWater
Resolved.

[JonathanM]

Try blkdiscard /dev/sdX at the console, where the X is replaced by whatever designation the drive happens to be assigned on that particular boot. That command will fully erase any content on an SSD, so be sure you have the correct /dev/sd designation.

[IceWater]

1 hour ago, jonathanm said:

Try blkdiscard /dev/sdX at the console, where the X is replaced by whatever designation the drive happens to be assigned on that particular boot. That command will fully erase any content on an SSD, so be sure you have the correct /dev/sd designation.

blkdiscard: cannot open /dev/sdb: Device or resource busy

 

[JorgeB]

Please post the diagnostics: Tools -> Diagnostics

[IceWater]

Here is the .zip, are you looking for a specific section? Happy to post that in text if so. 

Thank you

EDIT: The syslog had some personal info after browsing..

Here is what I found, if there is something else that could help please let me know...

 

: mkdir -p /mnt/cache
Apr 24 21:33:21 Tower emhttpd: shcmd (5553): mount -t xfs -o noatime /dev/sdb1 /mnt/cache
Apr 24 21:33:21 Tower kernel: XFS (sdb1): Invalid superblock magic number
Apr 24 21:33:21 Tower root: mount: /mnt/cache: wrong fs type, bad option, bad superblock on /dev/sdb1, missing codepage or helper program, or other error.
Apr 24 21:33:21 Tower emhttpd: shcmd (5553): exit status: 32
Apr 24 21:33:21 Tower emhttpd: /mnt/cache mount error: not mounted
Apr 24 21:33:21 Tower emhttpd: shcmd (5554): umount /mnt/cache
Apr 24 21:33:21 Tower root: umount: /mnt/cache: not mounted.
Apr 24 21:33:21 Tower emhttpd: shcmd (5554): exit status: 32
Apr 24 21:33:21 Tower emhttpd: shcmd (5555): rmdir /mnt/cache
Apr 24 21:33:22 Tower emhttpd: shcmd (5556): sync
Apr 24 21:33:22 Tower emhttpd: shcmd (5557): mkdir /mnt/user0
Apr 24 21:33:22 Tower emhttpd: shcmd (5558): /usr/local/sbin/shfs /mnt/user0 -disks 2 -o noatime,allow_other  |& logger
Apr 24 21:33:22 Tower emhttpd: shcmd (5559): mkdir /mnt/user
Apr 24 21:33:22 Tower emhttpd: shcmd (5560): /usr/local/sbin/shfs /mnt/user -disks 3 -o noatime,allow_other -o remember=0  |& logger
Apr 24 21:33:22 Tower emhttpd: shcmd (5562): /usr/local/sbin/update_cron

 

Edited April 25, 2021 by IceWater
RM log due to PII

[itimpi]

The whole zip is what is normally wanted.   However there appears to be something wrong as it cannot be downloaded.

[IceWater]

I was able to resolve this via pulling the drive out and running it in Windows Disk part with just the Clean command.
Honestly should of tried this sooner but I assumed that something similar was being done initially when you hit the "Format" button. 

Why is this not the case?

Also what would the Linux equivalent be to DIsk Parts  "clean" command which removes all volume and partition data and uninitialized? 

Thanks

[JorgeB]

17 hours ago, IceWater said:

Also what would the Linux equivalent be to DIsk Parts  "clean" command which removes all volume and partition data and uninitialized? 

 

On 4/25/2021 at 5:28 AM, IceWater said:

blkdiscard: cannot open /dev/sdb: Device or resource busy

This would completely wipe the device, without diags can't see why it didn't work, another option would be to use wipefs.

[Hassan.McKusick]

This is all bad advice and I appreciate trurl and itimpi for advising others to not follow in my footsteps. Turns out the root cause of most of my issues was that I needed to update my bios

I'm facing this issue now. Digging through the diagnostics the original issue seems to be that I used a tool to download files and the file names were too long. That borked my Docker btrfs and I've been having issues ever since. 

 As for the resource busy issue. When I run `lsblk /dev/sda` it points to `/dev/sda1` which points to `/boot`

 

I unmounted `/boot` with `umount -l /boot` but then I got stuck again. Some digging brought up a post on unix.stackexchange answered by ZakW that advised me to read the contents of `/proc/mounts` with `cat /proc/mounts`. Reading through `/proc/mounts` showed me that `/mnt/cache` is actually the mount point for my device and not the device "name" (not sure about that terminology).

`lsblk /dev/nvme0n1p1` pointed me back to `/mnt/cache` and with no other clues what to do I took a stab in the dark and unmounted `/mnt/cache`.

Finally I was able to run `blkdiscard /dev/nvme0n1p1`. I realized after the fact that I didn't run it on `/dev/sda` and once blkdiscard finished I tried running it on `/dev/sda` but I got the busy error again.

I rebooted my unraid box and when it came back up it showed the cache as missing a file system. On the "Main" tab I had the option to format the cache drive and everything looks good so far.

Edited July 5, 2023 by Hassan.McKusick
It's bad advice

[itimpi]

4 hours ago, Hassan.McKusick said:

I unmounted `/boot` with `umount -l /boot` but then I got stuck again.

I would expect this to cause problems    /boot is where the flash drive is mounted and Unraid expects to be available all the time it is running.

[trurl]

4 hours ago, Hassan.McKusick said:

When I run `lsblk /dev/sda` it points to `/dev/sda1` which points to `/boot`

 

I unmounted `/boot` with `umount -l /boot` but then I got stuck again. Some digging brought up a post on unix.stackexchange answered by ZakW that advised me to read the contents of `/proc/mounts` with `cat /proc/mounts`. Reading through `/proc/mounts` showed me that `/mnt/cache` is actually the mount point for my device and not the device "name" (not sure about that terminology).

`lsblk /dev/nvme0n1p1` pointed me back to `/mnt/cache` and with no other clues what to do I took a stab in the dark and unmounted `/mnt/cache`.

All of these command line attempts were not recommended and will lead to further problems.

 

Better to ask for advice in advance of doing anything like this.