* Sparse LUN scanning - 2.4.x
@ 2005-01-11 19:05 Jim Zajkowski
2005-01-12 0:07 ` Michael Clark
0 siblings, 1 reply; 4+ messages in thread
From: Jim Zajkowski @ 2005-01-11 19:05 UTC (permalink / raw)
To: linux-kernel
Hi there,
We have an Apple Xserve RAID, connected through a FC switch. The RAID
has LUN-masking enabled, such that one of our Linux boxes only gets LUN
1 and not LUN 0. We're running the 2.4.x kernel series now, since this
is under a RHEL envinronment.
The problem is this: since LUN 0 does not show up -- specifically, it
can't read the vendor or model informaton -- the kernel SCSI scan does
not match with the table to tell the kernel to do sparse LUN
scanning... so the RAID does not appear.
I can make the RAID show up by injecting a add-single-device to the
SCSI proc layer. Trivially patching scsi_scan.c to always do sparse
scanning works as well. No hokery with max_scsi_luns or ghost devices
works.
I'm considering making a patch to add a kernel option to force sparse
scanning. Is there a better way?
Thanks in advance,
--Jim
--
Jim Zajkowski OpenPGP 0x21135C3 http://www.jimz.net/pgp.asc
System Administrator 8A9E 1DDF 944D 83C3 AEAB 8F74 8697 A823 2113 5C53
UM Life Sciences Institute
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Sparse LUN scanning - 2.4.x
2005-01-11 19:05 Sparse LUN scanning - 2.4.x Jim Zajkowski
@ 2005-01-12 0:07 ` Michael Clark
2005-01-12 2:11 ` Jim Zajkowski
0 siblings, 1 reply; 4+ messages in thread
From: Michael Clark @ 2005-01-12 0:07 UTC (permalink / raw)
To: Jim Zajkowski; +Cc: linux-kernel
Jim Zajkowski wrote:
> Hi there,
>
> We have an Apple Xserve RAID, connected through a FC switch. The RAID
> has LUN-masking enabled, such that one of our Linux boxes only gets
> LUN 1 and not LUN 0. We're running the 2.4.x kernel series now, since
> this is under a RHEL envinronment.
>
> The problem is this: since LUN 0 does not show up -- specifically, it
> can't read the vendor or model informaton -- the kernel SCSI scan does
> not match with the table to tell the kernel to do sparse LUN
> scanning... so the RAID does not appear.
>
> I can make the RAID show up by injecting a add-single-device to the
> SCSI proc layer. Trivially patching scsi_scan.c to always do sparse
> scanning works as well. No hokery with max_scsi_luns or ghost devices
> works.
>
> I'm considering making a patch to add a kernel option to force sparse
> scanning. Is there a better way?
>
Add the Xserve with the BLIST_SPARSELUN flag into the blacklist/quirks
table in drivers/scsi/scsi_scan.c
~mc
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Sparse LUN scanning - 2.4.x
2005-01-12 0:07 ` Michael Clark
@ 2005-01-12 2:11 ` Jim Zajkowski
2005-01-12 9:50 ` Michael Clark
0 siblings, 1 reply; 4+ messages in thread
From: Jim Zajkowski @ 2005-01-12 2:11 UTC (permalink / raw)
To: linux-kernel
On 2005-01-11 19:07:53 -0500, Michael Clark <michael@metaparadigm.com> said:
>> The problem is this: since LUN 0 does not show up -- specifically, it
>> can't read the vendor or model informaton -- the kernel SCSI scan does
>> not match with the table to tell the kernel to do sparse LUN
>> scanning... so the RAID does not appear.
> Add the Xserve with the BLIST_SPARSELUN flag into the blacklist/quirks
> table in drivers/scsi/scsi_scan.c
It already is in the quirks list.
The problem is that LUN 0 does not show up on this machine, so the
quirks table doesn't work. Looking at /proc/scsi/scsi shows the device
but only sorta:
> Host: scsi1 Channel: 00 Id: 00 Lun: 00
> Vendor: Model: Rev:
> Type: Processor ANSI SCSI revision: ffffffff
whereas the LUN that is mapped shows up like this:
> Host: scsi1 Channel: 00 Id: 00 Lun: 01
> Vendor: APPLE Model: Xserve RAID Rev: 1.20
> Type: Direct-Access ANSI SCSI revision: 02
So since the information doesn't show up, the quirks table magic
doesn't work since it doesn't know that it needs to do a sparse lun
scan.
--Jim
--
Jim Zajkowski OpenPGP 0x21135C3 http://www.jimz.net/pgp.asc
System Administrator 8A9E 1DDF 944D 83C3 AEAB 8F74 8697 A823 2113 5C53
UM Life Sciences Institute
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Sparse LUN scanning - 2.4.x
2005-01-12 2:11 ` Jim Zajkowski
@ 2005-01-12 9:50 ` Michael Clark
0 siblings, 0 replies; 4+ messages in thread
From: Michael Clark @ 2005-01-12 9:50 UTC (permalink / raw)
To: Jim Zajkowski; +Cc: linux-kernel
Jim Zajkowski wrote:
> On 2005-01-11 19:07:53 -0500, Michael Clark <michael@metaparadigm.com>
> said:
>
>>> The problem is this: since LUN 0 does not show up -- specifically,
>>> it can't read the vendor or model informaton -- the kernel SCSI scan
>>> does not match with the table to tell the kernel to do sparse LUN
>>> scanning... so the RAID does not appear.
>>
>
>> Add the Xserve with the BLIST_SPARSELUN flag into the
>> blacklist/quirks table in drivers/scsi/scsi_scan.c
>
>
> It already is in the quirks list.
>
> The problem is that LUN 0 does not show up on this machine, so the
> quirks table doesn't work. Looking at /proc/scsi/scsi shows the
> device but only sorta:
>
Okay. I believe there is a patch floating around for 2.4 that is used in
some of the other distro's kernels that adds REPORT LUNS scanning as
well as a flag to force LUN scanning (although not sure it works if LUN
0 is not present).
A pragmatic solution may be to just add the echo scsi add-single-device
x x x x > /proc/scsi/scsi into your early boot or configure one of your
slices on LUN 0 (if the Xserve RAID allows that). 2.6 (and RHEL 4)
support the REPORT LUNS so should work there although there are some
messages on linux-scsi about the Xserve RAID's specific handling of LUN
0. Probably good idea to post to linux-scsi.
~mc
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2005-01-12 9:49 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-01-11 19:05 Sparse LUN scanning - 2.4.x Jim Zajkowski
2005-01-12 0:07 ` Michael Clark
2005-01-12 2:11 ` Jim Zajkowski
2005-01-12 9:50 ` Michael Clark
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).