I think this error will occur less frequently if the filesystem undergoes less stress. Any news?I have a new lead. I'm careful not to jump to conclusions since the issue seemed so random at first, but I now have a reproducible pattern. It may have been this pattern hiding from me in the past, or it may have been truly random, I'm not sure. But I've found that over the last days, the issue has always been triggered by my backup process, and specifically by the step in which I use rclone to sync my restic repository from the NFS mount to a remote B2 bucket. Today the process was stuck with a few files pending, and every time I tried to run rclone the transfer speed would instantly plummet to nearly 0, and the errors would start spamming my pi5's dmesg log. If I'm very lucky, I missed this before and the particular use case that fails is using rclone on the NFS mount. I'm in the process of moving the rclone step directly to the pi5, without NFS involved. We'll see in a few days if that solves the issue for good, or if I just removed an easy trigger but still face other triggers in the future.
Statistics: Posted by ejolson — Thu Feb 20, 2025 5:02 am