[SunHELP] redhat 5 / emc disk ?
listmail
listmail at triad.rr.com
Sun Jan 24 11:58:45 CST 2010
You mentioned DMX didn't you?
taken from powerlink:
----
spacer"Symmetrix DMX: Linux Server shows each LUN 9 times"
spacer
spacer spacer spacer
ID: emc154583
Usage: 12
Date Created: 02/12/2007
Last Modified: 01/07/2009
STATUS: Approved
Audience: Customer
Knowledgebase Solution
Question: What DMX-3 Director Bits do I need for a Linux Server with
QLogic Adapter?
Environment: OS: SuSE Linux Enterprise Server (SLES) 10
Environment: OS: Debian GNU/Linux, kernel 2.6.18-4-686
Environment: OS: Red Hat Enterprise Linux 5 (server)
Environment: Host adapter: QLogic QLA2312
Environment: Host adapter: QLogic QLA2340
Environment: Host adapter: QLogic QLA2342
Environment: Product: Symmetrix DMX-3
Environment: Product Symmetrix DMX-2
Environment: QLogic BIOS Settings "LUNs Per Target" = 256
Environment: scsi_mod Parameter max_luns = 255
Problem: Devices show up multiple times on invalid LUN Numbers
Problem: A Director Trace 91,30001 will show the QLA scanning all
LUN number up to 0xFFFF
Root Cause: Using the Linux standard FA setting, SCSI-2, the QLogic
Driver will scan all LUNs from 0(0x0000) to 65335(0xFFFF). As the 2
MSB(Most Significant Bits) do select the Addressing Mode and not
actually a LUN, the DMX will respond to these LUN Numbers with the
information of corresponding device. Additionally the DMX-3 will ignore
some bits in the FC LUN Address (e.g. Bit 5 in the LUN address filed of
the LUN Address Scheme) adding some extra Ghost Devices.
For LUN 0 this issue will show these LUNs:
* 0x0000 = 0x8000 = 0x8010 = 0x9000 = 0x9010 = 0xa000 = 0xa010 =
0xb000 = 0xb010
* 0 = 32768 = 32784 = 36864 = 36880 = 40960 = 40976 =
45056 = 45072
Once the SCSI-3 Flag is set, the Qlogic Driver will use the SCSI "REPORT
LUNS" command to discover the mapped LUN Addresses instead of scanning
each single Address.
That issue might occure on any other Linux distribution as well.
Fix: Method 1 (for SuSE Linux Enterprise Server 10 only):
A Novell kernel patch from 21.12.2006 did the job.
* patches.fixes/scsi-scan-limit-luns-seqscan-16k: Limit sequential
scan to 16k LUNs [#185164] 21.12.2006
* patches.fixes/scsi-scan-blist-update: Update blacklist entries for
EMC Symmetrix and HP EVA [#191648] 21.12.2006
Method 2 (not for RedHat 5):
As the scsi_mod parameter max_luns has no effect, you need to delete the
additional LUNs on the Linux Operating system. Any LUN above 255 could
be considered as additional. This issue should also show only additional
LUNs starting with 32768. You can delete any LUN with 4 or more digits.
* for i in
/sys/class/scsi_host/host*/device/target*:*:*/*:*:*:????*/ ; do
* echo 1 >$i/delete
* done
Method 3:
Enable the SCSI-3 Director Flag for the Linux Host. This can be achieved
using heterogenous on IBM_AIX.
----
listmail wrote:
> It's interesting to see something similar to your issue here
> https://www.redhat.com/archives/rhelv5-list/2009-January/msg00106.html
> about the 9 paths thing. They also tried powerpath and got 9 paths as
> well so maybe it's more likely to be in the hba driver config or
> kernel settings? I'm really not sure though without a box of my own
> to test with :) Could be a multitude of things. I actually
> understand the whole 'out of the box' preference but i'd still check
> what EMC certify would expect in case the answer is right there all
> along... Good luck.
>
> stephen price wrote:
>> Thanks for the responses - the unix team is pushing to use rh5
>> "out-of-the-box" with minimal adds/changes etc. So of course they
>> want to use the hba driver already bundled with the distro they
>> have. Same crowd complained when we (storage) told them they needed
>> to keep the emc drivers updated on sparc boxes in production &
>> development etc years ago. Not so much the worker bees as some
>> management etc.
>>
>> With the limited window that I'm going to have to work with this test
>> system - I will be testing the "real" emc driver, additional rpms
>> that benefit storage - fighting the selinux battle with netbackup
>> (selinux not certified with netbackup - unless it is "disabled") etc.
>>
>> So it appears to me that it will be a re-learning curve with rh5 -
>> which may be more painful than some of our management understands.
>>
>> Oh - here's the inq and lsscsi output - which is the current issue.
>>
>> Only a single lun (17) was presented - plus I see I'll have to figure
>> out on rh5 how to hide/blacklist the gatekeeper etc. More to the
>> to-do list <grin>.
>> #inq
>> /dev/sda :DELL :PERC 5/i :1.03 : : 142082048
>> /dev/sdb :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdc :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdd :Dell :Virtual Floppy :123 : : FAILED
>> /dev/sde :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdf :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdg :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdh :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdi :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdj :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdk :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdl :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdm :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdn :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdo :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdp :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdq :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sdr :EMC :SYMMETRIX :5773 :37001ad000 : 2880
>> /dev/sds :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>> /dev/sdt :EMC :SYMMETRIX :5773 :3700050000 : 35354880
>>
>> # lsscsi
>> [0:0:8:0] enclosu DP BACKPLANE 1.05 -
>> [0:2:0:0] disk DELL PERC 5/i 1.03 /dev/sda
>> [1:0:0:0] disk EMC SYMMETRIX 5773 /dev/sdb
>> [1:0:0:17] disk EMC SYMMETRIX 5773 /dev/sdc
>> [1:0:0:32768]disk EMC SYMMETRIX 5773 /dev/sde
>> [1:0:0:32784]disk EMC SYMMETRIX 5773 /dev/sdf
>> [1:0:0:33025]disk EMC SYMMETRIX 5773 /dev/sdg
>> [1:0:0:33041]disk EMC SYMMETRIX 5773 /dev/sdh
>> [1:0:0:36864]disk EMC SYMMETRIX 5773 /dev/sdi
>> [1:0:0:36880]disk EMC SYMMETRIX 5773 /dev/sdj
>> [1:0:0:37121]disk EMC SYMMETRIX 5773 /dev/sdk
>> [1:0:0:37137]disk EMC SYMMETRIX 5773 /dev/sdl
>> [1:0:0:40960]disk EMC SYMMETRIX 5773 /dev/sdm
>> [1:0:0:40976]disk EMC SYMMETRIX 5773 /dev/sdn
>> [1:0:0:41217]disk EMC SYMMETRIX 5773 /dev/sdo
>> [1:0:0:41233]disk EMC SYMMETRIX 5773 /dev/sdp
>> [1:0:0:45056]disk EMC SYMMETRIX 5773 /dev/sdq
>> [1:0:0:45072]disk EMC SYMMETRIX 5773 /dev/sdr
>> [1:0:0:45313]disk EMC SYMMETRIX 5773 /dev/sds
>> [1:0:0:45329]disk EMC SYMMETRIX 5773 /dev/sdt
>> [2:0:0:0] cd/dvd Dell Virtual CDROM 123 /dev/scd0
>> [3:0:0:0] disk Dell Virtual Floppy 123 /dev/sdd
>>
>>
>> regards
>> steve
>> _______________________________________________
>> SunHELP maillist - SunHELP at sunhelp.org
>> http://www.sunhelp.org/mailman/listinfo/sunhelp
> _______________________________________________
> SunHELP maillist - SunHELP at sunhelp.org
> http://www.sunhelp.org/mailman/listinfo/sunhelp
More information about the SunHELP
mailing list