[smartmontools-support] Problem while running long device test

[Christian Franke]

David C. Partridge wrote:
> I have an Adapted 51245 Raid card, I installed a new device and defined a
> simple volume on that device, and then started a long smart test:
>
>   smartctl -t long -C -d aacraid,0,0,7 /dev/sdc
>
> this started fine, but some few hours later I got this in syslog:
>
> Nov 16 21:25:06 Charon smartd[852]: Device: /dev/sdc [aacraid_disk_00_00_7]
> [SCSI], failed to read SMART values
> ...
> Nov 16 21:25:07 Charon smartd[852]: Device: /dev/sdc [aacraid_disk_00_00_7]
> [SCSI], failed to read Temperature
> ...
> Nov 16 21:25:42 Charon kernel: [16757.128862] aacraid: Outstanding commands
> on (1,0,1,0):
> Nov 16 21:25:42 Charon kernel: [16757.148889] aacraid: Host adapter reset
> request. SCSI hang ?
> ...
> dev sdb, sector 0
> Nov 16 21:26:27 Charon kernel: [16802.517265] Buffer I/O error on dev sdb,
> logical block 0, async page read
> Nov 16 21:27:13 Charon kernel: [16848.025581] aacraid: Host adapter reset
> request. SCSI hang ?
> Nov 16 21:27:28 Charon kernel: [16863.389743] aacraid: Host adapter reset
> request. SCSI hang ?
> Nov 16 21:27:28 Charon kernel: [16863.389757] aacraid 0000:01:00.0: Adapter
> health - 217
> ...
>
> The system was then hung.  I'm wondering if the problem occurred because the
> test was still running in foreground mode and that the attempt to read smart
> data was blocked as a result.

Very old info from smartctl man page (actually unchanged since very 
first release 2002):
"WARNING: Tests run in captive mode may busy out the drive for the 
length of the test. Only run captive tests on drives without any mounted 
partitions!"

Which implies than any attempt to access the drive during the test will 
fail. The syslog suggests that the failing accesses from smartd 
triggered host adapter resets which failed again and again because this 
drive is virtually bricked during the test.

Why do you need a captive test here ?

Regards,
Christian