I noticed my vm (Ubuntu 14 server) running on my cache pool drives suddenly became read only.  Low and behold one of my cache drives appears to be failing with reallocated sectors.  I'm currently running the scrub operation on the cache pool (Should I even be doing this?)  When done can I replace the failing disk and start a rebuild operation? Is there anything else I should be doing?  Thanks. 

I noticed my vm (Ubuntu 14 server) running on my cache pool drives suddenly became read only.  Low and behold one of my cache drives appears to be failing with reallocated sectors.  I'm currently running the scrub operation on the cache pool (Should I even be doing this?)  When done can I replace the failing disk and start a rebuild operation? Is there anything else I should be doing?  Thanks.

 

Firstly, I am sorry that I am answering this post with an I don't know. But honestly, in answer to all your questions, I don't know.

 

I am posting because I'd like an answer to the questions myself. I think there is a very BIG gap in documentation re BTRFS Cache Pools with respect to recovery, rebuild, expected behaviour and guidance.

 

Jonp seems to the foremost authority on this sort of thing BUT he has been conspicuously absent (as have ALL the LT Staff) for some time. Presumably they are preparing v6.2 release BUT who knows? Communication does tend to dip to zero when new releases are due.

 

In the absence of me being able to help you, consider this post an acknowledgement that someone has read your post and understands your issue and at very least provides a bump to your thread, issue and questions.

 

Would should always attach the diagnostics.zip to a support post, without it I’ll be doing some guessing.

 

If one of the disks is bad you should remove it from the pool, scrub can’t fix bad sectors, if there are any, and if there are it can cause more harm that good, note that simply unassigning it is not enough, you have to physically disconnect it or cache pool will appear as unmountable when trying to start the array.

 

Start array with the good cache disk and see if all works, you can also do a btrfs check:

 

Edit: array has to be started in maintenance mode to run btrfs check

 

btrfs check /dev/sdX1

 

replace X with your cache letter.

 

 

If there are errors you can try and repair them:

 

btrfs check --repair /dev/sdX1

 

If all is good with the single disk you can just add another one to the pool, balance (rebuild) will automatically be done when you start the array.

 

Thanks for the replies.  Fortunately I have a spare disk.  When I replace the failed disk, will Unraid automatically format as BTRFS?  I had prviously precleared it and formatted it as XFS.

If you add it to the cache pool it will be formatted btfrs, you should never have to format it manually.