[SOLVED] Docker service died. Cause: directory and shares issue.

[Xenu]

Earlier today my Unraid (along with Docker services) became unaccessible from a web browser.

 

I gracefully powered down my machine and powered it back on. I was then able to access Unraid through my browser again, but the Docker service couldn’t start up. The given error was that the 'Docker vDisk location' and 'Default appdata storage location' didn't exist. According to the settings page these were set to /mnt/user/system/docker/docker.img and /mnt/user/appdata/. When I FTP'd into my server I noticed these directories did in fact not exist, but they did if I changed user in the directory for cache. This brought the Docker service back online, however, every Docker container is behaving like it was freshly installed. I've lost all progress and data made with each container.

 

Looking at the each containers configuation, its still looking for its config/appdata files in the /mnt/user/appdata/, not the /mnt/cache/appdata/ directory. The original config/appdata files still exist, but under the new 'cache' directory.

 

Why did this error occur? I never changed the directory settings for the Docker service.

How can I get back all the progress/data associated with each container?

How can I prevent this happening again?

Is the cache directory the correct location for this docker.img and appdata directory?

box-diagnostics-20201202-1350.zip

Edited December 2, 2020 by Xenu

[Squid]

Dec  2 12:57:20 box emhttpd: shcmd (46): /usr/local/sbin/shfs /mnt/user -disks 15 2048000000 -o noatime,allow_other -o remember=root  |& logger
Dec  2 12:57:20 box shfs: fuse: invalid parameter in option `remember=root'
Dec  2 12:57:20 box shfs: fuse_main exit: 3

This is your entire problem here, so effectively it appears that the entire /mnt/user/.. is unavailable.

 

A quick scan of other diagnostics I've downloaded show remember=some number, and not "root"  Did you manually change a config file on the flash?

[Xenu]

10 minutes ago, Squid said:

Dec  2 12:57:20 box emhttpd: shcmd (46): /usr/local/sbin/shfs /mnt/user -disks 15 2048000000 -o noatime,allow_other -o remember=root  |& logger
Dec  2 12:57:20 box shfs: fuse: invalid parameter in option `remember=root'
Dec  2 12:57:20 box shfs: fuse_main exit: 3

This is your entire problem here, so effectively it appears that the entire /mnt/user/.. is unavailable.

 

A quick scan of other diagnostics I've downloaded show remember=some number, and not "root"  Did you manually change a config file on the flash?

Thank you for the reply.

 

The only users on my Unraid is the default 'root' user and 'James'.

 

I never changed any directory or config settings. This change of directory settings and unaccessibility literally happened by itself in the earlier hours of the morning.

 

Upon further investigation I've noticed whilst all my containers are still looking for their appdata and config files in /mnt/user/appdata/ they have all been moved over to /mnt/cache/appdata/.

Edited December 2, 2020 by Xenu

[trurl]

Cache is part of user shares and appdata user share is usually on cache.

 

Nothing got moved but your user shares are all broken. 

[Xenu]

8 minutes ago, trurl said:

Cache is part of user shares and appdata user share is usually on cache.

 

Nothing got moved but your user shares are all broken. 

Just looked, definitely an issues with the shares as you said. The shares page on Unraid is empty/showing no shares.

Any idea what might've caused this and how I can resolve this whole issue, please?

Edited December 2, 2020 by Xenu

[JorgeB]

You need to remove "root" from Settings -> NFS -> Tunable (fuse_remember): leave it blank to use default value.

[Xenu]

9 minutes ago, JorgeB said:

You need to remove "root" from Settings -> NFS -> Tunable (fuse_remember): leave it blank to use default value.

Thank you for the reply!!! Wow, it was as simple as that!

 

Your reply reminded me I was following a guide online a little while before the issue and I may have changed that field to "root". I noticed no negative impacts following so forgot. I have now tried clearing the field and applying the change, though it kept changing itself back to root, after repeating the process few more times it changed to '330.' Following a reboot a populated shares page has been restored and after changing the Docker services directories back to how they were previously, everything is working again.

 

I really appreciate your help! Thank you!!!!!!!!!

[Ghawkserver]

On 12/2/2020 at 7:36 AM, JorgeB said:

You need to remove "root" from Settings -> NFS -> Tunable (fuse_remember): leave it blank to use default value.

I'm new to using Linux and Docker in Unraid and I am having the same issue where my shares are empty and I cannot figure a way to save them properly or create a share to allow docker to startup and start to run at all. Can you offer any help? I've tried to watch all the videos on YouTube and all the how to forums but this is the fist thing I've seen in the realm of help. Thoughts? Thank you!

[JorgeB]

11 minutes ago, Ghawkserver said:

Thoughts?

Please post the diagnostics.

[Breathtaker]

Hello together,

 

unfortunately I am facing the same issue and it is not happening for the first time. First time it occured before the update to 6.11.5 and I had to edit the docker path to "/mnt/cache/appdata/" to bring the containers back to life. After the reboot of the update, everything was back working normal.

 

This time I turned off my Unraid server, to add more RAM. After it was booted on and I wanted to install a docker instance of MariaDB, i was getting an error.  As i checked the path i could see that the default path for "appdata" is again no more available.

 

Please find attached the diagnostics file.

 

Edit:

I rebooted the server again and now "appdata" is back again.

 

Thank you.

cube-diagnostics-20221229-1400.zip

Edited December 29, 2022 by Breathtaker