Since some weeks I am experiencing a random behaviour with one of my external mounted SSD disks. I have a server with 3 disks in total (1 nvme where nixos os is installed and 2 SSDs mounted on my home). All 3 disks are luks encrypted and btrfs formatted.
I am not really sure about the culprit but I have noticed that sometimes when I start a big file transfer from my Windows laptop to the server by accessing the disk folder via Samba, transfer can become very slow or kind of stops if I had previously stopped a previous samba transfer.
Sometimes this resolves but in other cases I have noticed that CPU usage (not ram) starts to increase a lot by looking at htop. nbmd application seems to be on top of cpu usage. By killing that service via systemctl-tui, cpu usage still slowly increases (see picture below) and ssd disk where transfer was interrupted becomes not responsive. For example, giving an ‘ls’ command does not return anything back. I have to force the reboot the server to get access to that disk again.
All the 3 disks have scrubbing enabled and no errors have been found.
Do you know how I could find out what is going on?