To All,

We have a mixed environment of users accessing the system using Terminal Server (Citrix Metaframe XP) and PCs. All PCs are Windows 2000. OI and Data sits on a file and printer server. We have setup a dedicated indexer with a user INDEXER.

The system was installed last week. When it works, it works great. But we are seeing slowing down after certain number of users are logged in. For example, if 6 people are logged in, things are working fine. However, if the 7th individual tries to log in OI really slows down and gives the appearance of being "locked". Users can be accessing the system from a thin-client or Win2K PC.

Also, for those that can login in to work, OI seems to perform fine for a period of time then slows down, then picks up speed again. Needless to say this is annoying.

When the user tries to exit out of a form, they may get the windows "Not Responding" message. Periodically, the user may even get "kicked out" of OI32 completely. This also doesn't make a good impression.

For the NT Services, we are using Named Pipes and TCP/IP. For dedicated indexing, the background time was set to 10 seconds. I increased this to 300 seconds and this doesn't seem to help. (Should I go lower instead??)

Any suggestions or comment for getting OI to perform better. Hardware should not be an issue as we are using dual Xeon processors for the File and Print Server and for the Terminal Servers. They also have fast SCSI drives 15Krpm and have 4GB of memory.

Are there specific tuning issues for Citrix Metaframe XP and W2K?

HELP!!

Ray Chan

We have seem a slow down on TS/Citrix during logon processes.

It is caused by the non-unique station id.

Set the station Id using a stored procedure registered against your application environment. There are many posts on this topic, but pls advise if you need help.

We append the current time to the station id to make it unique, it is a very simple three line program.

We have seen this improve speed immediately.

In your case though you state that the speed is slowing down whilst actuall using the system, which to me is a slightly different problem that I am suggesting above.

On one of our Citrix sites, I did see this, and the problem then was a mixture of hardware and setup. One of the network cards was a bit 'iffy' and there was a problem with the connection between the Citrix server and the application/data server. I would suggest that this is a good place to check.

Best of luck

Colin

Colin,

Thanks for your response.

At the moment, we believe that the "glitch" is with the W2K service using TCP/IP in our mixed enviornment of Terminal Server and PC clients. Hardware problem is not an issue.

Briefly, we can now duplicate the symptons that our client sees (slow slow slow down - appears locked, breaks to debugger). We can do this by putting the LH service into debugger mode and put it into "Pause mode". When you do this every workstation is unable to "click" on anything in OI. If you were to stop the LH service while our OI app is running it will break to debugger. This mirror very closely the "lost of connection" sense that our users are experiencing. This leads us to believe that something may not be working correctly with the LH service in our situation. Something may be causing it to "hiccup".

The two options we are faced with is to either: 1) turn off the service or 2) change the communication method to Named Pipes.

We will first change things to Named Pipes and see if that will help. If not, we will TURN OFF the service. Needless to say, having a system that cannot run continuously is causing our clients to doubt us, but they have been relatively patient, but it will not last too much longer so I must find a solution soon. Tomorrow (January 2, 2003) the client office will be reopen for business. That will be the real test.

Well thanks for the thought and yes wish us luck,

Ray Chan

Glad you have the problem area pinpointed, but as a matter of interest do you have the unique station id facility in place?

Ray - guessing here…

Have you checked for the network cards being configured to use burst mode? Try toggling this setting in the device setup in Windows / on the card, and see if it helps. Also, is the slowdown due to one (faulty or mis-configured) network card operating at 10mbit when all other cards on the same leg are at 100mbit?

Steve

The Sprezzatura Group

World Leaders in all things RevSoft

Steve,

I can take a look at this. Clarification: Should the card be in "burst mode" or should in be in some other mode?

Thanks

Rah

Ray - I'd tend to set the cards *slow* (non-burst) and see if this improves things.

Steve

The Sprezzatura Group

World Leaders in all things RevSoft

We are currently using OI 3.7.5 and noticed the same issue. Our issue is that the NTLM procces is killing our dual 1GB servers. I am still looking for a solution to this problem. NT4 TSE seemed to handle the issue fine, but that is also a 16 bit OS, not a 32 bit OS like Win2k. If anyone has any solutions, please let me know.

View this thread on the forum...