Problem! - Intermittent Connectivity 2

[xiong]

I'm having a problem on an internal LAN with my users intermittently receiving the following error:

Cannot connect to the Citrix server:
The Citrix Server you have selected can not be located

There have been no changes to the system in a few days, and then it was only an installation of Windows 2000 SP3. There were no issues immediatly following the SP install.

This is happening to all of my internal and external clients and is happening 65% of the time.

Info:
MetaFrame XPa FR2 - One Server
Windows 2000 Pro SP3 Clients
External clients connect thru PIX firewall
Internal clients connect directly
IIS 5 running on Citrix box, NFuse
20 CALs, 4 in use


Any thoughts?

 

[xiong]

...In addition...

At all times, every user can enter the Citrix Program Neighborhood and refresh the desktop. They can connect, enter their login credentials, and the icon set will refresh appropriatly.

 

[xiong]

Solution:

As it turns out, a corrupted DNS entry on one of our servers that points to the Citrix server was corrupted and needed to be recreated. Upon recreation, everything seems to be working as it should again.

 

[xiong]

Strike that. Although everyone was immediatly able to connect... a few short hours later, we were back to the same issues.

Restarting the DNS did not help.

Stopping the DNS service and restarting on a new server did not help

 

[CitrixEngineer]

At the point in time you get this message, can you ping the Citrix server? By name?

The fact that you found corrupt records on your DNS server does suggest that it may have a problem.

Are all your clients using the same DNS server in their IP stacks?

As a workaround, you could try inserting the name or IP of the Citrix server into each of your client's IP stacks - although that's a little tedious.

There's no timeout setting inherent in MetaFrame that would cause this, so continue troubleshooting your network and client IP settings. This message comes from the client at a point at which communication with the server has been lost. If it was the Citrix server, then 100% of your users would have the same issue simultaneously.


Hope this helps CitrixEngineer@yahoo.co.uk

 

[xiong]

Thanks but no luck. I went the extent of stopping the DNS server altogether, and still had the same issues.

I am able to ping the server, and open shares on that server, even though the server returns a "not found" message.

 

[xiong]

In addition.

Most users can open an ICA desktop, but not individual applications.

 

[xiong]

More Info...

After stopping the IMA service and recreating the LHC, all local users (inside the firewall) were able to connect. When an external user attempted to open an application, the problem started again with 80% failure to connect to the server.

 

[jdl508]

Can you have the remote user add an entry to their host file locally to them. Can the remote user ping the server no problems by name? Another thing could be possibly going through a router the packet could get fragmented if to large I had this happen to remote users at my site and needed to set a dont fragment rule on my router. I'm not sure if the same applies to your FW.
jdl

 

[xiong]

The remote users have entries in their HOSTS file that redirect them to the server. At all times they can ping that server, open web-pages on that web-server, FTP to that server, etc... In fact, all users can have one published app open, and will still get the "Server not Found" error when attempting to open a second app.

 

[30362]

Have you tried re-installing any MF service pack since the Win2k-SP3 update? I had a similar problem with a MF1.8-SP3, re-appying SP3 fixed it.

I am not familiar with MFXP&FR2.

Have you gone so far as to try and do NETMON traces of client connectivity to the server, the trace log files might help shed some light if it is ICA, DNS etc..

Also, do you still have the RDP protocol enabled?

 

[xiong]

I haven't tried a reinstall of FR2 for MetaFram, but not a bad idea.

RDP is still enabled.

I have not run NETMON, but I'm not sure that would shed any light on the subject.

Right now we're leaning towards a corrput LHC and DS databases, corrupting on every connection. It seems at this point, that the first user after recreating both of these, is able to connect just fine. After that, it's a give or take.

 

[xiong]

Thanks for all the help, but I've found the solution.

A lot of the troubleshooting lead to an indication that the Data Store was getting corrupted after only a few users made a connection. So it made sense that an ICA desktop could still be opened, as it did not need to access the DS.

After reviewing a few hundred articles in the Citrix Knowledgebase, I came across one that related to SQL and Oracle Data Store configurations. One of the minor notes was relating to a problem tha might occur if a new Load Management definition was created on a non-load managed server. Evidently in these scenarios, the IMA may lose connectivity to the Data Store.

Granted, I work in a Data Store that is Access-based, and the server is not load managed. But during some "self-training" I had created a new definition a few days back. I removed that definition, and after that, there were no connectivity issues. Why it took a few days to rear its ugly head, I don't know, but this worked.

SOLUTION:
Remove all Load-Managed definitions that are not created during installation, for servers that are not load-managed.

 

[CitrixEngineer]

Thanks for posting your solution, eslersa - too often people find a solution to an issue after posting, but don't tell us how they fixed it!

That could be really helpful to someone else, so I'm giving you a star CitrixEngineer@yahoo.co.uk

 

[benbro62]

Hi
I've a problem with my citrix farm. I've 10 w2k/sp2 and Metaframe Xpe sp3/fr3 servers upgraded fron MF 1.8/sp3 and my DS is a dedicated SQL2000 server.
All work fine just i've not problem with SQL and DS.
When i stopped SQL server all clients are not able to connect at Public Application or directly at the Desktop of any server in the farm.
After i restart the sql server all connection returns to work fine. What's happen. I know that if DS stop the farm work good for netx 48 hour, It's not my case, it's work for 10 seconds!!!!!!!!!!!

Please I need urgently any suggestions.

regards

Franco

 

[PamMcC]

similar problem, xiong touched on it early in the post. Ran a sniffer against attempted tcp/ip+http connection that received i/o error. Ica resolved fine in dns - client contacted citrix server just fine and vice versa. but when the server pointed the client to a server on which to launch the connection...i/o error. 2 of our 19 servers did not have entries in dns. Added them back and no more i/o error.

 

[tk36]

I had the similar problem, I did the search and I found your answer. Thank you so much, xiong.