HI All,
I'm having a couple issues with our Orion NPM / VoIP / APM / NTM machine. Before I go into the problems, here's the necessary vitals:
Machine Vitals:
- Win2k3 Server SP2
- 2.8Ghz dual core Xeon
- 3.00 GB RAM
- SQL Database is located on a SQL server across the network, not local to the server
Orion Vitals:
- Orion 9.5.0 May 2009
- NTA 3.1.0.0
- VoIP 2.1
My VoIP plugin is acting very weird (imo). I have 19 locations connected via MPLS and I have call-paths configured as full mesh. Yes, I understand this is ALOT of SLAs for the plugin to process. But, I've noticed on my routers (mostly cisco 2801s [12.4(22)T], a couple 4507Rs) I am seeing this message every 5 min:
Jul 15 10:18:40.124: %SYS-5-CONFIG_I: Configured from 192.168.101.200 by snmp
Jul 15 10:07:58.364: %SYS-5-CONFIG_I: Configured from 192.168.101.200 by snmp
Jul 15 10:13:17.012: %SYS-5-CONFIG_I: Configured from 192.168.101.200 by snmp
That is the IP of my ORION server. I can only think that it is Orion configuring the SLAs. Orion isn't set to do any other sort of configuring. Why does the VoIP Monitor continuously reconfigure these SLAs? Shouldn't it be able to configure the SLAs once and be done with it?
Also,
On the "Voip Sites overview", it's almost always saying "Enter Read/Write credentials In Orion.With Orion v9 to populate/edit community strings you can also use ...." for each host. This doesn't make any sense, the nodes have the correct snmp settings (using snmp v3). If I go to Admin -> Manage Nodes -> Edit Node, i can hit the "Verify SNMP" button and it will be successful.
Also, the VoIP Monitor is polling data from the hosts, its displaying it, so it can obviously access the device. The VoIP monitor is uses 95% total CPU for 30-120 seconds per polling cycle. I've tried restarting it, restarting the server, but the thing just hammers away at the server's CPU. Any thoughts why this happens? I wouldn't imagine processesing the results from 170 snmp polls would be all THAT CPU intensive.
Another problem, I consistently see: Error "Object reference not set to an instance of an object." occurred during polling VoIP Call Path Node1->Node2 (simulation node Node1 [192.168.x.x])" but the nodes are different. It might be an error referencing Node 10 talking to Node 4 or somthing, there's no pattern. The only thing I can think of is that the box is being overwhelmed with the task of creating / analyzing the SNMP queries. I typically see one to two of these errors every polling cycle.
The problems don't seem to coincide with any updates. It has always been this way. I've applied Orion / plugin updates in hopes that it would resolve some of these issues.
I've tried increasing the time between pollings, but that hasn't helped. Is there somthing awry with the VoIP Monitor? Or do you think the server is simply being overloaded? Any help would be apreciated. Thanks.
Summary of problems:
- Recreating IP SLAs on routers
- High CPU utilization
- Errors reading SNMP credentials
- Object reference errors.
