smartd fails to report disk failure if a disk doesn't respond anymore

[kaluscha]

I had a self test running on disk hdb:

smartd: Device: /dev/hdb, self-test in progress, 10% remaining

The disk encountered problems, see /var/log/messages:

kernel: hdb: dma_timer_expiry: dma status == 0x61
kernel: hdb: DMA timeout error
kernel: hdb: dma timeout error: status=0xd0 { Busy }
kernel: ide: failed opcode was: unknown
kernel: hda: DMA disabled
kernel: hdb: DMA disabled
kernel: ide0: reset: success

There were several kernel IDE resets until the drive didn't respond anymore:

kernel: hdb: drive not ready for command

smartd wrote messages:

smartd: Device: /dev/hdb, failed to read Temperature

However, smartd had been configured to send e-mails in case of trouble (/dev/hdb -a -I 194 -W 4,40,42 -R 5 -m myamil). In this case, it failed to do so.

In my opinion this is a major problem as smartd should inform the admins that a disk is complety offline, i.,e. doesn't respond to requests on the IDE bus anymore.

[Christian Franke]

smartd sends a warning email "failed to read SMART Attribute Data" in the above situation, see smartd.cpp​. No additional email "failed to read Temperature" is sent because temperature info is part of the attribute data.

If smart option -s, --savestates is used see also ticket #35.