[Scst-devel] Scsi command errors

[Johan K. <joh...@ca...>]

Hi! I=B4m new to this list, but known by some of you from iet list. First =
I=20
want to thank Vlad for his efforts, this is a nice piece of software, =
and=20
we hope that we will be able to contribute to the developement of it. =
We=20
are going to use scst in a bigger project, that involves iet, evms, and=20
some other software. We have a professional storage setup, with many=20
oportunities. At the moment, we have setup like this: First a workbench =
as=20
a host, running Win2k, with a QLA 2202 HBA connected to a SANbox =
switch.=20
The switch is connected to QLA 2200 target HBA on Storior, our =
experimental=20
storage appliance, which runs Lineox Enterprise Linux 4.0(a Red Hat =
clone),=20
kernel 2.6.14. Storior is then connected through an LSI 7202 initiator =
HBA=20
to another SANbox switch, and that switch is finally connected to an IBM =
Fc=20
ProFibre 1100J JBOD storage array, with 5 st318452fc seagate fc drives.=20
Through scst we exported two drives ask disk devices, and they showed =
up=20
nicely in the workbench HBA bios setup list of devices. When we went in =
to=20
the HBA setup, we where able to verify data and media. It was also=20
functioning very well to make volumes and format them in windows, and =
after=20
that to move data around. No problems with some quick tests. But when =
we=20
tried to low level format, through the workbench HBA setup tool, we got=20
scsi command errors, like:
scsi operation code: 04
scsi sense key: 06
scsi additional sense code: 29
scsi additional sense code qualifier: 00

After this, the drives were corrupt, and it was impossible to access =
them=20
from the workbench, and all the way to the LSI 7202 HBA on Storior. In =
the=20
Storior LSI 7202 HBA setup tool, you could not access the drives with =
the=20
verifying tool, so you have to low level format them, which went =
allright.=20
Now the drives are functional again.

Of coarse we want to know what was happening here. A theory might be =
that=20
the QLA HBA didn=B4t work very nice together with the LSI HBA, when it =
sended=20
the low level format command. JK