I have a 50 TB BTRFS filesystem on a hardware RAID array in an HPC cluster I manage. Users have filled up this partition to 100% and complained about I/O errors while reading old files. The filesystem went read-only.
I unmounted the partition. It took a while to unmount and it finished with "segmentation fault". Then I tried to mount it back and it is already taking 20h with no sign of progress.
In the past it used to take extraordinary time (5-10min) to mount it, but this time it is orders of magnitude more and no guarantee it would ever finish.
Do you have suggestions how to proceed? Would fsck on a full filesystem work?
