S.M.A.R.T. smart long selftest aborted by host

[Douglas Tutty <dtutty@porchlight.ca>]

Hello,

I have two Seagate Baracuda 7200 80 GB drives.  I've got smartmontools
and have smartd set to monitor all attribues but I don't have it run
self-checks automatically.  The computer isn't on all the time so I
can't schedule in advance when it will be on.  I'd like to check the
drives manually.  Note: I don't think there have been any problems, but
I'd like to know how to do it before I have a problem.  

Everything works fine for sdb:

	/usr/sbin/smartctl -t long -d ata /dev/sdb

then check status with:
	
	/usr/sbin/smartctl -a -d ata /dev/sdb

will show me the progress and eventually that it "completed without
error".

When I do the same with /dev/sdb, it gets to 90% remaining and shows
"Aborted by host".

I thought that perhaps the smartd perodic polling of smart data was
confusing it (so to speak) so I:
	/etc/init.d/smartmontools stop

then reran -t long... with the same result.

Any ideas on how I get /dev/sda to complete a long test?

Thanks,

Doug.

----
INFO:

Neither drive likes the -o on option either (get syslog error
about setting auto offline testing failed).

Each drive has three partitions: first ones are raid1 for /boot, second
ones are raid1 for LVM VG system, third are not raid but straight LVs
for VG local (right now, just for /home).

mdadm shows that the drives are sync'd ok and shutdown -rF shows
filesystems are clean.