Hi,
On 06/19/2014 06:49 PM, Veeraraghavan, Kugesh wrote:
+ Martin who can provide assistance going forward.
-----Original Message-----
From: Netapp-Linux-Community
[mailto:[email protected]] On Behalf Of Sarraf,
Ritesh
Sent: Thursday, June 19, 2014 10:08 PM
To: Hans van Kranenburg; [email protected]
Cc: Sarraf, Ritesh; Bill MacAllister; [email protected]
Subject: Re: [Netapp-Linux-Community] Bug#740701: multipath-tools: mkfs fails "Add.
Sense: Incompatible medium installed"
Hans,
SCSI UNMAP functionality got complete very recently in the Linux kernel. From
what I see so far, you seem to be running on 3.2 kernel. The Debian kernel
team's policy for stable is to only backport important fixes, and some device
driver refreshes. I highly doubt they'd have backported SCSI enhancements.
If UNMAP is important to you as a feature, you may want to try evaluate a more
recent kernel.
I just checked on b.d.o, it currently has Linux 3.12. That should be a good
start to verify against.
https://packages.debian.org/search?keywords=linux-image&searchon=names§ion=all&suite=wheezy-backports
As long as I have not succeeded to construct a proper isolated
reproducible test case which fails consistently, there's not much reason
to start changing or trying anything I guess.
So, right now I have built a copy of the production environment setup
where this occured (see prev messages in debian bug report [1]), and I
cannot trigger the same bug there yet, so something important must still
be different. Same linux kernel, same ONTAP version, same layering of
iscsi, multipath, dm_crypt and lvm...
In the meanwhile, I've been browsing around a bit in the kernel git
history, also comparing latest (3.16 now) and stable linux-3.2.y.
When taking the error messages as starting point...
sd 12:0:0:0: [sdc] Unhandled sense code
sd 12:0:0:0: [sdc] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
sd 12:0:0:0: [sdc] Sense Key : Medium Error [current]
sd 12:0:0:0: [sdc] Add. Sense: Incompatible medium installed
sd 12:0:0:0: [sdc] CDB: Unmap/Read sub-channel: 42 00 00 00 00 00 00 00
18 00
end_request: I/O error, dev sdc, sector 4372576896
...I see that the only place in the linux kernel where the message
"Unhandled sense code" occurs is in drivers/scsi/scsi_lib.c, around line
900, where a case statement is executed on the sense key value.
SENSE KEYS are defined in drivers/include/scsi/scsi.h. One of them is
MEDIUM_ERROR, 0x03, which seems to be the one we got here. MEDIUM_ERROR
is not handled in the case statement, so the fall-through is executed:
897 default:
898 description = "Unhandled sense code";
899 action = ACTION_FAIL;
900 break;
This is the case in 3.2 and in 3.16 as well. There are changes related
to UNMAP or WRITE_SAME in this area, but only in case of
ILLEGAL_REQUEST. One of those changes (66a651a) is related to preventing
complete path failures on failed operations. But, that codepath is not
chosen in this case anyway, so not relevant.
I really want to understand what this message means and who generated
it. If I assume it's the NetApp filer that returns this data when
issuing an UNMAP command, I wonder why. Could any of you NetApp folks
shed some light on when and how ontap 8.1.2 7-Mode might ever send this
error back? This could help finding a test case which will trigger this
error, and to find out why it does not occur in a different situation
that seems to be identical.
The whole error still does not make a huge amount of sense to me. Why
would my NetApp system return an error which I only would expect to see
when using a CD-ROM drive or some USB removable device? It's not like
all disks suddenly vanished from my FAS and disk shelves, like ejecting
a CD drive. :-)
[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=740701
--
Hans van Kranenburg - System / Network Engineer
T +31 (0)10 2760434 | [email protected] | www.mendix.com
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
