Hi Alan,
It sounds suspiciously like one of two issues.
1) Memory leak or exhaustion
2) SMB 1.0 issues
Issue 1 can be addressed on both W2K and W2K3 by following Microsoft TechNet article KB312362. By default, the Memory Manager tries to trim allocated paged pool memory when the system reaches 80 percent of the total paged pool. Depending on the system configuration, a possible maximum paged pool memory on a computer can be 343MB (Windows 2003 Standard) and 80 percent of this number is 274MB. If the Memory Manager is unable to trim fast enough to keep up with the demand, then you may receive event ID 2020 (The server was unable to allocate from the system paged pool because the pool was empty). But that is just one symptom.
By tuning the Memory Manager to start the trimming process earlier (for example, when it reaches 60 percent), it would be possible to keep up with the paged pool demand during sudden peak usage, and avoid running out of paged pool memory.
Microsoft will tell you that setting these values should be considered as a best practice, especially for busy servers. The optimum percentage for trimming will vary, but the default recommended setting (60%) is a good place to start.
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management]
"PagedPoolSize"=dword:ffffffff
"PoolUsageMaximum"=dword:0000003c
If the paged pool memory was running close to the edge, something like a new AV engine/pattern or MS Hotfix could have pushed it over. You can use Microsoft’s poolmon to monitor this further.
Issue 2 can be “managed” by monitoring some perfmon counters and then tuning out the SMB parameters as per the articles listed here:
http://www.jhouseconsulting.com/index.php/blog/2008/05/15/smb-tuning/
If either or both the paged pool trimming and/or SMB is really busy, it may eventually recover over time, which goes close to perhaps explaining why it sometimes fixes itself without a reboot.
Either way, MetaFrame XP is no longer a supported product and now 4 major product releases behind.
Cheers,
Kind regards,
Jeremy Saunders
Senior Solution Architect - Virtualisation Specialist | Datacom Systems WA | 29 Oxford Close, West Leederville, WA 6007 Australia
Email: jeremy.saunders@datacom.com.au | Ph: +61-8-9210-0806 | Mob: +61-413-441-846 | Fax: +61-8-9380-4226
Personal Blog: http://www.jhouseconsulting.com/index.php
From: thin-bounce@freelists.org [mailto:thin-bounce@freelists.org] On Behalf Of alan tropper
Sent: Thursday, November 13, 2008 1:43 PM
To: thin@freelists.org
Subject: [THIN] Meta Xpe 1.0 servers showing as down in the CMC
Hi,
I hope someone can help me, we have 20 citrix servers in our farm and they are mixed 2000 & 2003 with a couple on VM. All servers have Meta Xpe 1.0 with latest feature release and service pack.
However some of the servers drop off the farm during the day and also at night, once re-booted they come back into the farm and show as up and then it happens again.
I have tried to re-build the problem servers LHC file but the issue is still happening, the servers processor and page file don’t seem to be too busy at the point when the servers drop off resource manager. The servers themselves stop allowing new users to log on but still maintain current connections so this points to a timeout somewhere with maybe the metric or zone server not getting a response from one of our servers in time….although why do they not come back into the farm straight away, occasionally they will come back as up without a re-boot but we cant carry on working like this in the environment.
The CMC does throw up an error code: 527 when try to select server and in the event logs we do see a Perflib event ID 1015 timeout error on the servers and event ID 257 which refers to Resource Manager unable to retrieve metrics. All servers are on the same subnet and can be pinged from and to each other
I have seen an article on increasing a Terminal Server timeout session but not how to actually do it…any help would be greatly appreciated!
Thanks in advance
Al
Alan Tropper
Server Operations
Customer Service Centre
Information and Communication Technoligies
Department of Education and Training | 151 Royal Street, East Perth WA 6004
Tel: (08) 9264 8126 or 5555 | Fax: (08) 9264 4701 | alan.tropper@det.wa.edu.au
This document is intended solely for the named addressee. The information contained in the pages is confidential and contains legally privileged information. If you are not the addressee indicated in this message (or responsible for delivery of the message to such person), you may not copy or deliver this message to anyone, and you should destroy this message and kindly notify the sender by reply email. Confidentiality and legal privilege are not waived or lost by reason of mistaken delivery to you.
No comments:
Post a Comment