Lookout 6.0 issue - Remote Hypertrends and nicitdl5 CPU Utilization

[erblock]

Issue Symptom:

Very High server CPU utilization (nicitdl5.exe) during remote client sessions that use Hypertrend or NIHypertrend (Active-X control) objects, each/either configured with multiple traces.

Client processes with Hypertrends running on the same machine (server, in my case) containing the Citadel 5 database (also the machine that runs server processes writing to the database…in my case) are not affected with these symptoms.

 

Note: I could not get even half this (4400 character count attachment) to post without a "Message cannot exceed 5000 characters" error.  So I attached the original full message.

[erblock]

Hmmm.... No one wants to touch this one?  Please open and read the attachment to get the full post.  No Viruses or nasties....I promise.

Ed Block

MIT, Cambridge, MA

Note for Webmaster or anyone at NI that cares!...For some reason,  text pasted from an MS Word document with a character count of ~4400, wouldn't go up.  Even ~2500 characters (about half the attachment text) wouldn't go up either.  Are there some characters that these forums wont interpret correctly, or is the forum message size interpretation software just messed up?

[Ricardo_SR]

Hello,

I will pass the report about the DF problem.  About the high CPU in version 6.0, it is something that we are aware of it and we are working to improve the performance.  The reason for the high CPU is that Citadel 5.0 is much more robust than Citadel 4 and there is a lot more of error checking.

The performance should be improved soon.  We are working on it.

Ricardo S.

National Instruments

[erblock]

Thanks for the Acknowledgement Ricardo!

As for error checking and robustness being the issue at hand....I do'nt think so!  If that were the case, the problem would show with more data being taken out overall, not more data being sent to one trend object.  Afterall, single dicreet traces on many (20 for example) individual trend objects in a process uses 10 times less nicitdl5.exe CPU resources than a single trend with only 8 discreet feeds???

It seems to me that a data-trace in a database is just that.  It shouldn't take more resources to send many traces to one remote object than it takes to send twice as many traces to many different remote objects...right??

A more detailed explaination would be nice...but of course a timely fix is the bottom line.

There are bugs in there....BIG ones.  And....ones that have been there for a long time since release.

 I'll get working on the other issues I've discovered with Lookout 6 to see if those are "known" as well.

Of course I still love the product "line" overall, but as is, Lookout 6 is currently out of the question... totaly unusable!

Anyway....

Cheers and Thanks Again. 

Ed

[erblock]

I have now also verified that a process web client will also cause the symptom on the server.

Any time frame for a fix?

[Mike@DTSI]

ive been waiting for a long time myself... Search under my name.

[skibum019]

I too have been waiting for improvements on 6.0 and have also been greatly disappointed with the Citadel 5.0 database.  Many of the same problems you all are seeing we are experiencing as well.

[MDN]

I too have had this issue with Lookout 6.0. The current "fix" I've been given by NI tech support is to downgrade to Lookout 5.1. So far that has been the only working solution for the problem. I've had no contact from NI regarding this issue for the past month. All I can conclude is that Lookout 6.0 shipped with a critical flaw that has yet to be resolved.

My advice to those who are considering upgrading to Lookout 6.0 is to wait until the next version comes out and has been demonstrated to be stable.

[erblock]

Good!

Come on Lookout users....Lets keep the pressure on...maybe we will get some results.

Maybe even get some official NI update correspondence on this and the other LK 6.0 Hypertrend and nicitdl5.exe issues...  Soon???

Gotta hope anyway....

[Mike@DTSI]

Even after talking to multiple Engineers at NI on the phone.  Was told they MAY ship me some "testing code" to see if it helps, havent hear a thing.  We have (1) system of the 5 new these last few months running 6.0.  More than just database issues. 

I love the simplcity of Lookout, while still being powerful.  But if things keep this way, we may have to look into other SCADA software... 

BTW, i also do e-Commerce systems using SQL server, well over 300 concurrent users on a database and have yet to see usage of the processor as high as Lookout does...

Keep us informed!!