Wednesday, June 10, 2009

[THIN] Re: Disk Queue Length Question

From: thin-bounce@freelists.org [mailto:thin-bounce@freelists.org] On Behalf Of Saravanan Srinivasan
Subject: [THIN] Re: Disk Queue Length Question

 

 

Did you test with local profile user settings?

No, the customer rebuilt the VM and now all the before mentioned counters are staying below 1.  "Something" must have been FUBARed in the original VM.  No users are reporting errors or frozen sessions.  If all goes well this week they will probably get corp mgmt  approval to proceed with the rollout and move to XenServer.

 

Thanks

 

 

Webster


--- On Mon, 6/8/09, Webster <Webster@CarlWebster.com> wrote:


From: Webster <Webster@CarlWebster.com>
Date: Monday, June 8, 2009, 12:24 PM

From: thin-bounce@freelists.org [mailto:thin-bounce@freelists.org] On Behalf Of Joe Shonk
Subject: [THIN] Re: Disk Queue Length Question

 

How many users are on this box?  How much memory/cpu did you give each of the VMs?  Are you also redirecting the Application Data folder as well? Or is it part of the roaming profile (cached locally)?

 

There are 20 users on this box.

 

Each VM has 4GB RAM and 1 vCPU.

 

AppData is NOT redirected (I listed to Shawn at Synergy!).  It is part of the Roaming Profile cached locally.

 

The customer is in the 2nd part of testing for a major rollout.  They had no issues with their initial 5 users.  When they went to 20 users all hell broke loose.  Their goal is to support 1,500 users – 1,200 concurrent.  Once they get the kinks worked out for these 20 users, they will move to XenServer 5 and their new Del R900 servers.

 

 

Webster

 

 

From: thin-bounce@freelists.org [mailto:thin-bounce@freelists.org] On Behalf Of Webster
Subject: [THIN] Disk Queue Length Question

 

Greetings,

 

Have a customer having an issue with frozen sessions on a TS2003/XA5 x86server.  The issue happens with both RDP and ICA.  And happens remotely or logged in to the console.

 

Issue: When using Adobe Acrobat 9.1 Pro to create PDFs from various form of documents, the sessions freeze for very long periods of time.  i.e. > 20 minutes

 

Info: TS/XA server is a VM under Hyper-V.  My Docs and Roaming Profiles are redirected to File Servers that are also VMs under Hyper-V.

My Docs and Roaming Profiles are stored on an EMC SAN connected via 4GB FC.  Both are configured for RAID5 as that is the only option available to them until they do a major upgrade on the SAN sometime this summer.

 

Problem: Avg Disk Queue Length is > 1,300

Avg Disk Read > 4,500

Avg Disk Write > 22,000

 

Question: Of course I know that taking, on avg, 22 SECONDS to write a block of data is going to cause MAJOR issues and is more than likely the cause for the session hangs/freezes.  But, on a FC SAN, are Avg Disk Queue Length and or Current Disk Queue Length valid counters to monitor for a Terminal Server/XenApp Server?

I am finding conflicting advice on whether the Disk Queue Length counters are valid for either SANs or Roaming Profiles and redirected My Docs stored on a SAN.

 

No comments: