On Sun, 22 Jan 2017, Jan Vlach wrote: > I would like to report kernel panic on Sun Fire V245 running OpenBSD > 6.0-stable (Kernel is GENERIC.MP with erratas up to 15, minus 11 > (xenocara))). > > The panic message is: "softdep_deallocate_dependencies: unrecovered I/O > error"
Your dmesg shows this: > /: got error 5 while accessing filesystem > panic: softdep_deallocate_dependencies: unrecovered I/O error "error 5" is EIO, a physical I/O error. That almost certainly means that there was an unrecoverable bad block on the disk, in the area for the root partition. Get a new drive, copy the data over and toss this one. > I've tried to force the coredump too, but last boot shows that there was > no core to save into /var/crash. I've probably messed-up somewhere. Well, the dump *also* hit a bad block: > dumping to dev 7,1 offset 8388604 > dump /data: got error 5 while accessing filesystem ...so the suggestion is really strong to save your data while you can and replace the drive. > The server comes up back just fine with hwraid intact and passes auto > fsck repairs. > > Is this machine starting to show it's age or is this an actual bug? fsck doesn't read (or read+write) every block, it over checks the structural data...but I would suggest you not poke at the drive until you have any and all data you want off of it. Philip Guenther
