Losing hypertrend data on remote computers in Lookout 6.1

[JPhillips]

I know this issue has occurred in the past, but because we ares still losing trends too often,  I am at the point where I am going to have to create expressions on my client process file for every data member I want to have on a hypertrend so that I can have each client machine get its trend data from a local database rather than pulling the data from the database on the server. Before beginning this tedious task, I wanted to get some input on why this might be happening.

When we lose trends I have to restart the server in order for the trends to come back. At first I thought is was caused by nicitdl5.exe crashing but I went into the settings for this service and set it to restart everytime it crashes. This did not fix the problem.

My client process file contains approximately 250-300 hypertrends with 24hr. time periods trending both numeric and logical data members (water levels and pump run statuses). Could the number of trends I have in my process file could be causing the issue (too many)?

Also, when I lose trends I have opened NIMax to try and create a trace for one of the members being trended and I cannot view the trace either - (no data) is displayed in the trace description.

And, another strange thing that occurs is immediately after closing my server process file, the trends appear on the client machine before I restart the server process file.

Some things I have done to try and correct the problem are:

1.) reduced the resolution for the numerics from .001 to 0.01.
2.) the resolution for the logicals from .001 (default) to 0.1.
3.) reduced the lifespan from perpetual to 15 days (did that today).

My server machine is a Dell PWS690 with 3.25 GB of RAM running Windows XP Pro, Service Pack 2, and the lookout version I am running is 6.1 (build 3)

Due to the size and location of our client computers, they have to go through 2-3 subnets to get to our server where the database resides.

Could this be a network issue where Lookout for whatever reason is disallowing access to the database after a period of time or because of a networking "hiccup"?

Plus, would using a local database to to display the info on these trends be a viable option in your opinion?

Please note: I AM NOT losing communication with my server as far as information coming from my plcs and being transmitted to the clients from the server.

Thanks for any quick response.

[Ryan.S]

Firstly, one thing I want to confirm with you.

You said you didn't see the trace in MAX when the trends were lost, on server computer or client computer? Can you see the trace in MAX on server computer?

[JPhillips]

No, I cannot see the trace on the server. That is the only computer that I use MAX on 99% of the time.

[erblock]

The first thing I would check/try is to set the clients for Administrator login with no expiration.

Next, make sure the user logged on at the client, has read permissions on the database (Citadel) directory of the server.

Next, make sure the client firewall is allowing the necessary NI executables.

Make sure your client symbolic links point to machine names that are always resolvable to the correct IP address.  For instance if you have two separate dns paths or a dual homed server machine (2- nics on different networks), this can cause the behavior.  For this case you could use a lmhosts file to fix the problem possibly.

Any/all of the above can cause the issue, but the only one that would likely cause the issue to come and go is the first one I think (lookout login).

[JPhillips]

Thanks for the info, erblock.

Related to first comment, are you saying to add the client users to my administrator group on my lookout server machine?

[erblock]

Maybe Ryan and crew can clear this up....

All that is needed for the discreet object connection form a remote client to the server is an IP address of the server and firewall ports on the server to be properly opened.  There are references online for the Firewall settings.

Hypertrend connections are a different animal and I would like NI to step in and explain this if possible:

What I have observed is that the traces must be using a commonly blocked (by ISP's) port.  I have to use a VPN to get the traces to work from an internet client.  I also must either be logged on to the domain of the server machine or map a local network drive to the server Citadel (Database) directory.  The map only needs read permissions.

I haven't been able to find any documentation online explaining the differences between live oject data (Logos?) and trace data connections (also Logos?...but different somhow?) over the Internet with Lookout and/or DSC and I would like to understand this better myself.

Ed

Message Edited by erblock on 04-01-2008 09:29 AM

[JPhillips]

I finally resolved this problem by adding a batch file to my scheduled tasks that kills nicitdl5.exe every hour on my server, and since I have nicitdl5.exe set to restart after failure, the service automatically restarts.

I have been running lookout this way for approximately 3 weeks and have not had an issue of trends disapperaing on my clients any more.