Failure of mirror disk causes /dev/sdas to become read-only

Latest response

I have a situation where one disk in a hardware raid1 configuration has failed and now the disk device is marked as read-only. Any idea why this would happen? I would think that the raid controller should hide the failed disk from the OS. Looks like there was a 20 seconds timeout failure after ResetTarget which caused the offlining.. Any ideas how to get the disk online again without a reboot?

hdparm -I /dev/sdas returns
/dev/sdas: No such device or address

The controller is a HP Dynamic Smart Array B140i Controller setup as RAID 1/RAID 1+0

Here are the relevant portions of /var/log/messages

Oct 7 11:50:35 obsora101 kernel: hpdsa: PAL_VM_ResetTarget.
Oct 7 11:50:46 obsora101 kernel: hpdsa: PAL_VM_ResetTarget.
Oct 7 11:51:06 obsora101 kernel: sd 3:0:0:0: [sdas] Medium access timeout failure. Offlining disk!
Oct 7 11:51:06 obsora101 kernel: sd 3:0:0:0: Device offlined - not ready after error recovery
Oct 7 11:51:06 obsora101 kernel: sd 3:0:0:0: [sdas] FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_TIMEOUT
Oct 7 11:51:06 obsora101 kernel: sd 3:0:0:0: [sdas] CDB: Write(10) 2a 00 05 efb4 e0 00 00 10 00
Oct 7 11:51:06 obsora101 kernel: blk_update_request: I/O error, dev sdas, sector 99595488
Oct 7 11:51:06 obsora101 kernel: EXT4-fs warning (device dm-7): ext4_end_bio:332: I/O error -5 writing to inode 57 (offset 101306368 size 4096 starting block 473244)
Oct 7 11:51:06 obsora101 kernel: Buffer I/O error on device dm-7, logical block 473244
Oct 7 11:51:06 obsora101 kernel: Buffer I/O error on device dm-7, logical block 473245

CF

Responses