Eval running, service randomly stops

Yes, the service should be on and not stop unless there is a problem.  We currently have a bug that has a workaround. 

Could you look in the NT Event Log to see the error message that is comming from the NetFlow service?  It is under the Control Panel\Administrative Tools\Event Viewer.

Thanks,
David

Hello, can you please post the workaround?  I just started running NetFlow in Orion again, after addressing some other db issues.  The NetFlow service crashed last night.  The error from the Event Viewer on my system is:

Event Type:    Error
Event Source:    NetFlowService
Event Category:    None
Event ID:    0
Date:        5/4/2009
Time:        11:37:43 PM
User:        N/A
Computer:    xxxxxx
Description:
Critical error in NetFlow listener: Exception of type 'System.OutOfMemoryException' was thrown.

This occurred with NetFlow 3.1  I just installed sp2 this morning and have NetFlow running again to test (although the webpage still just says "NetFlow 3.1" and does not indicate sp2??).

After searching out other threads here, I found a suggestion that the db may not be able to keep up, therefore the NetFlow process on the polling engine keeps consuming more and more memory until it overflows/errors.

Not positive this is what is occurring in my environment but it could be.  The avg disk queue length on the db server takes a big hit with NetFlow going.  I am looking to upgrade that server but in the meantime would like to keep NetFlow running if this workaround is applicable to my case.

*edit* Also meant to add that in our environment we are talking about 10's of thousands of flows.  I haven't figured out how to determine the exact number of flows Orion is handling yet.

Also, the NetFlowService and the w3wp.exe service seem to climb in memory utilization pretty close to each other.  Probably related.

Hello, the service crashed again last night.  Should I open a case with support?  I am fairly certain the underlying cause is that the current db server disks cannot keep up, therefore the NetFlow service on the polling engine (separate systems) continuously consumes more memory until it overflows.  Shouldn't it simply truncate when it reaches this critical point, instead of crashing?

Yes, sorry about the delay.  The event log error that you are receiving is not like the issue that we've seen and have a workaround for.  Please enter a case with support.

Thanks,

David

David,

Case #94908 opened was opened on May 12th.

Do you know if there has been any progress?  I have made NTA work by limiting the number of flows pointed at it.  Currently, it is working with only ~1,500 PDU's average.  When I point flows from my entire core network at it (~9,000 PDU's avg), the NetFlowService.exe process continually consumes more memory and crashes within 24 - 72 hours.

If you guys need any further information, let me know.  I'd be happy to help in whatever way possible.

I've just pinged Tech Support to find out where things are at with your case.  Did they send you a hotfix to install?  We now have one for customers with your situation.  I'll try to make sure that they contact you soon.

Thanks,
David

Thank you!  Just got an email from them.  I appreciate the work all of you put into this.  :)  Not sure if this also addresses the OP's issue, otherwise I'd suggest it as a solution.