random disconnects and performance, in the Citrix XenApp / Presentation Server forum on BrianMadden.com

Hi,

When you setup the test CPS 4.5 farm on a different Internet connection, were you still going through the PIX/ASA? This sounds like a network infrastructure issue.

Check your interface stats on the PIX and any intermediary switches (hopefully they are manageable) and look for resets, CRCs, etc. Make sure you have all your speed/duplex settings correct and make sure that you don't have an MTU problem either.

Cheers,

Have you checked if there are any problems in comunicating with domaincontrollers?
Have you checked were all network traffic is going to?
From there on, you can eliminate this as a problem or start to work on it...
My experience was that a domaincontroller on another site was validating citrix users and sessions started to freeze.

The second internet connection does go through a pix. I have checked all interfaces for speed/duplex issues and all of the interfaces as well. Nothing there. No errors, no resets...nada. As far as the MTU issue, not really sure how to check for that.

for my DC, I can communicate to it from my citrix box if that is what you are asking.

Hi,

The simplest way to check your max MTU size is to use PING with the -l and -f switches as described here:

http://support.microsoft.com/kb/314825

The -f switch sets the DF bit high on the packets, so that all the hops know not to fragment the packet. The -l switch tells PING how big of an ICMP packet buffer to make. Make sure you PING a remote server (i.e. www.google.com)

Set -l to 1472 and see if you get this error message:

"Packet needs to be fragmented but DF set"

If so, drop the value of the -l switch by 10 (i.e. 1462) and PING again. Keep dropping the value until packets go through properly. That's your maxMTU.

If the maxMTU is any value lower than 1472, you'll need to adjust your MTU on your PIX (and perhaps elsewhere) as pathMTU discovery isn't working and large packets are getting dropped. Alternatively, you can try to figure out why (usually due to ACLs on routers, often at ISP level).

Cheers,