NO problem, if it took me 6 hrs with 3 tier 3 support engineers at
Microsoft Exchange, Performance, and a Lead Performance Engineer to
troubleshoot the root cause of the issue, its definitely a FYI for
everyone out there dealing with SAN's and not just EMC's either Hitachi
has been mentioned, you saw the Symantec Post, there are probably others
too, they I am not privy to since I don't use there Multipath software,
but being an EMC shop it tends to lend credence accordingly that this is
what the root of the cause is. 

 

Z

 

Edward Ziots

Network Engineer

Lifespan Organization

MCSE,MCSA,MCP+I, ME, CCA, Security +, Network +

[email protected]

Phone:401-639-3505

________________________________

From: Troy Meyer [mailto:[email protected]] 
Sent: Tuesday, June 30, 2009 9:20 AM
To: NT System Admin Issues
Subject: RE: For those that use multipathing solutions with your servers
NON Paged memory pool leak

 

WOW

 

Thanks for the post Ed.  In a previous life I spent hours
troubleshooting this issue and never came to this conclusion.  I have
forwarded your links on to the guys that are at my old gig and I am sure
they will appreciate it.

 

-troy

 

From: Ziots, Edward [mailto:[email protected]] 
Sent: Tuesday, June 30, 2009 4:34 AM
To: NT System Admin Issues
Subject: For those that use multipathing solutions with your servers NON
Paged memory pool leak

 

All, 

 

http://seer.entsupport.symantec.com/docs/317378.htm 

http://blogs.technet.com/andym/archive/2008/12/04/powerpath-5-2-and-mpio
-causing-npp-leak.aspx

 

The culprit in EMC land is the 5.2 and 5.2 SP1 PowerPath Drivers (
MPIO.sys of 1.22.3790.2358) and the EMCMPIO.sys of 5.2.1.6 (5.2 SP1)

 

The fix: Move up to EMC multipath drivers Version 5.3 which will load
Microsoft's MPIO.sys updated to 1.23.3790.2451 which seems for the time
being to address the Non-Page Memory Pool leak, this was driving one of
my Exchange 2003 SP2 2-node clusters insane over the last few months, it
was discovered if the NON-Paged Memory counter you see on the Task
manager gets about 106-108MB for a X86 32bit system, the http service
for the Exchange Cluster will fail, and cause the Exchange group to fail
because the resource is marked to affect the group. 

 

HTH with some folks out there, 

 

Z

 

 

Edward Ziots

Network Engineer

Lifespan Organization

MCSE,MCSA,MCP+I, ME, CCA, Security +, Network +

[email protected]

Phone:401-639-3505

 

 

 

 

 

~ Finally, powerful endpoint security that ISN'T a resource hog! ~
~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/>  ~

Reply via email to