Pool is deactivated once all LUNs' paths lost, no tolerance

[osl_maohuang]

Dear all,

I have a test case disabling all paths and enabling them 30s later on switch. I assumed the pool on NW6.5 SP8 should wait a moment, at least 60, and then be deactivated. Actually the pool is deactivated immediately, about 8s later.
I have tuned the timeout of QlogicHBA and SCSI by below parameters on STARTUP.NCF:

LOAD QL2X00.HAM SLOT=201 /LUNS /ALLPATHS /PORTNAMES /XRETRY=12 /XTIMEOUT=120
LOAD QL2X00.HAM SLOT=202 /LUNS /ALLPATHS /PORTNAMES /XRETRY=12 /XTIMEOUT=120

LOAD SCSIHD.CDM /RETRIES=8 (which means SCSI timeout is 40+8*11=128seconds)

Could any body throw any hint on this issue?
Thanks in advance.

BTW, I have enabled Multipath and its timeout is 60.

[JdelaFerte]

Hi,

Sounds to me that it is working as designed..
If I understand you correctly you have 2 paths to the storage and you disable both paths an re-enable them again after 30 secs?

I think you're mixing two things here, one is the Qlogic HBA/SCSI and a complete other thing, NSS.
The pool is being deactivated to prevent corruption, as it lost it's connection to the storage, now you can set your SCSI timeout as high as you want, but NSS will kick in anyway and disabled the pool to prevent further damage as read and write will fail.
This should stroke with what you see on the server as well, as you will probably see a lot of 20203 and 20204 and also 20206 errors (NSS read and write errors and disabling pool)

Hope this help and if I complete mis understood you then also let me know 🙂

Cheers,
Johan

[osl_maohuang]

jdlf_wf4;2000799 wrote:
Hi,

Sounds to me that it is working as designed..
If I understand you correctly you have 2 paths to the storage and you disable both paths an re-enable them again after 30 secs?

I think you're mixing two things here, one is the Qlogic HBA/SCSI and a complete other thing, NSS.
The pool is being deactivated to prevent corruption, as it lost it's connection to the storage, now you can set your SCSI timeout as high as you want, but NSS will kick in anyway and disabled the pool to prevent further damage as read and write will fail.
This should stroke with what you see on the server as well, as you will probably see a lot of 20203 and 20204 and also 20206 errors (NSS read and write errors and disabling pool)

Hope this help and if I complete mis understood you then also let me know 🙂

Cheers,
Johan

Hi Johan,
Thanks for your reply.
You've fully got me. I am fresh on Netware and attempted to test the tolerance of NSS pool when it lost all paths to storage device.
So, is there any parameter to enable/disable this NSS feature? or it's not tunable?

[JdelaFerte]

I'm not aware of such an option in NSS, furthermore, why would you have a timeout of 30 secs on a pool if this pool lost its storage?
If a pool disables itself it is for a good reason and tries to prevent (more) damage.

I would focus on tuning the multipath solution, that is what your pool depends on.

Cheers,
Johan

[mrosen]

Hi,

osl maohuang wrote:
> You've fully got me. I am fresh on Netware and attempted to test the
> tolerance of NSS pool when it lost all paths to storage device.

That tolerance in Netware is basically zero, for good reasons. Netware
values data integrity higher than *anything* else.

> So, is there any parameter to enable/disable this NSS feature? or it's
> not tunable?

Nope.

CU,
--
Massimo Rosen
Novell Product Support Forum Sysop
No emails please!
http://www.cfc-it.de

[osl_maohuang]

I got it. Thanks, Johan & Rosen.