Terminal services/citrix random freezes

[localzuk]

I have a random thing happening on my terminal services boxes where the sessions randomly lag by a few seconds. I've enabled disk caching as suggested by various bods, and i've fixed everything that was being thrown up in the event viewer so can't see what's causing it.

The servers are highly specced, with 4GB (well, 3.4GB after windows messes with it all), dual 2.4Ghz quad core xeon's, and dual 143GB SAS 10k rpm hard drives, and gigabit connections.

This problem occurs even when there are only 1 or 2 people logged on.

So, any suggestions would be much appreciated!!

[CyberNerd]

sounds very much like a network problem to me.
do you have a network baseline that you can compare with?

[Axel]

Does this happen with both thin-clients and PCs running RDP....?

[localzuk]

The same problem happens with RDP sessions also.

How would I identify a network issue? The switches are no-where near even half utilisation. The network utilisation on the server is also minimal (ie. maxing out at about 4%, averaging 0.7%!), even when it has 20 clients attached.

[CyberNerd]

Ideally you already have a baseline from which to compare....
Use (something like) iperf SourceForge.net: Iperf to measure your network performance.

You could probably do a quick check by doing an aggressive ping (sudo ping -f target) on a *nix. this will quickly show up if any packets are being dropped.

[localzuk]

After doing a ping -f on the server for 10 minutes, I can report zero packets lost.

However, when I do a wireshark capture on the server, most of the ica packets are coming up as checksum incorrect (possibly due to checksum offloading). Should I disable checksum offloading?

[CyberNerd]

I just googled and found this:
[TCP CHECKSUM INCORRECT] : checksum, tcp, incorrect

[localzuk]

That seems to say 2 things... ie. 'OMG! That's awful' and 'No worries, that's just Checksum offloading'.

I may disable it and see if it solves this issue.