Docker Service failed to start and more...

[Luneski]

Hi everyone,

 

A month ago, my unraid server started to behave strangely. My docker would stop "working" randomly and when I go to Docker tab on the GUI, some docker would be stopped and some wouldn't. If I try to touch anything, Unraid would throw an error that I did not note. As my planning was quite tight, I only did a reboot and everything was back on track. But the issue reappeared about one time a week...

 

Since yesterday, the issue evolve to its final form...

 

Since this appeared, I cannot get docker to start again. I got my eyes on some logs but I'm no expert and don't find anything relevant.

 

I went to Docker settings to try to delete the Docker vDisk image but the page refresh instantly and nothing is done. I also tried to go manually delete de vDisk image which is located in /mnt/user/system/docker/docker.img but I get an error that says :

root@Tower:/mnt/user/system/docker# rm docker.img 
rm: cannot remove 'docker.img': Read-only file system

 

And since then I'm stuck, so could you please have a look at my daig files and maybe head me somewhere I will find the solution ?

 

Thanks in advance.

 

Luneski.

tower-diagnostics-20220729-1353.zip

[Squid]

Jul 29 11:20:25 Tower kernel: BTRFS info (device nvme0n1p1): bdev /dev/nvme0n1p1 errs: wr 0, rd 0, flush 0, corrupt 609, gen 0

The cache drive has detected corruption and mounted itself as Read only to attempt to protect the system.  You should start by running a Memtest from the boot menu for at least a pass or two to eliminate the possibility that memory errors caused this (if you're booting the OS via UEFI, you will need to temporarily switch it to legacy boot in order for Memtest to work)

 

As an aside, I'm not personally a fan of using BTRFS for a cache pool if you have no intensions of expanding that pool to be a multiple device.  XFS is more forgiving on things like unclean shutdowns, memory errors etc.

 

@JorgeB

[Luneski]

Ok, maybe I can completely unmount the device and put a new one in place ( I've got like 10 nvme drives not used), but what will I lose if I do so ? I'm not going to use anymore than one drive as cache so I'll go and switch to XFS as per your advice.

[trurl]

memtest is the first thing

[trurl]

1 minute ago, trurl said:

memtest is the first thing

and no evidence the drive is failing

[Luneski]

Ok I'll gove a go at the memtest when I'll be physically next to the server.