I recently changed from a single-SSD XFS cache to a dual SSD BTRFS cache. I stopped VM's and Docker, moved everything off the cache drive, reformatted it to BTRFS and added a second one, then moved everything back. I deleted my docker.img file and redownloaded all docker images from the "previous apps" tab as has been mentioned many times on this forum.

 

I must have done something wrong because when i try to install a new docker container i now get the following error:

 

Error: open /var/lib/docker/tmp/GetImageBlob484538067: read-only file system.

 

I do not know where to start solving this. My diagnostics file is attached to this post. Many thanks in advance for any help.

tower-diagnostics-20201020-1526.zip

Edited October 20, 20205 yr by YourNightmar3

Pool fs is fully allocated, you need to run a balance, see here for more info, king of strange this is happening in a new filesystem, did you fill it up completely before?

8 minutes ago, JorgeB said:

Pool fs is fully allocated, you need to run a balance, see here for more info, king of strange this is happening in a new filesystem, did you fill it up completely before?

Thank you for the quick reply.

Now that i think about it, i did accidentally set a share that wasn't supposed to be on cache to cache and it filled up the cache completely. I didn't think much of it as i put the share cache usage back to normal and ran mover again.

 

I ran

btrfs balance start -dusage=75 /mnt/cache

with the following result:

Done, had to relocate 203 out of 934 chunks

But unfortunately when creating a docker container i am still getting "Error: open /var/lib/docker/tmp/GetImageBlob091507751: read-only file system"

Edited October 20, 20205 yr by YourNightmar3

2 minutes ago, YourNightmar3 said:

But unfortunately when creating a docker container i am still getting "Error: open /var/lib/docker/tmp/GetImageBlob091507751: read-only file system"

You'll need to re-start the docker service, that should do it, if it doesn't recreate the docker image.

8 minutes ago, JorgeB said:

You'll need to re-start the docker service, that should do it, if it doesn't recreate the docker image.

Thanks so much! It worked!

 

I did find that thread you linked when googling for the problem but i also found some people saying i had to reformat my entire SSD cache so im glad it was solved with this solution instead.

 

Why is this command nessecary when the cache fills up? Will i have to do this again in the future?

This should no longer happen with recent kernels, but in some circumstances it does, you're not the first.

Thanks alot. My NVME cache filled up and I got same error after this.

On 10/20/2020 at 10:00 AM, YourNightmar3 said:

Thank you for the quick reply.

Now that i think about it, i did accidentally set a share that wasn't supposed to be on cache to cache and it filled up the cache completely. I didn't think much of it as i put the share cache usage back to normal and ran mover again.

 

I ran

btrfs balance start -dusage=75 /mnt/cache

with the following result:

Done, had to relocate 203 out of 934 chunks

But unfortunately when creating a docker container i am still getting "Error: open /var/lib/docker/tmp/GetImageBlob091507751: read-only file system"

i think i am chasing a similar issue but not certain but when i tried to run the balance cmd i got the error 

ERROR: error during balancing '/mnt/cache': Read-only file system

 

pepper-diagnostics-20210413-0751.zip

36 minutes ago, Cakroksh said:

ERROR: error during balancing '/mnt/cache': Read-only file system

Cache filesystem is corrupt, you should backup and re-format.