rsfh

That's what they gave me when I called in the problem. You can refrence my case #102141 if it'll help. Post back if you notice any problems after implementing the fix and I'll do the same. Hopefully this time it'll work out. 

rsfh-- What version of NPM are you currently running? M NPM 10.1.3 APM 4.0.2 SP4

I was having the same problem. NPM SL500 9.1 SP5 & APM ALX v2.5. Running on Server 2003 with SQL 2005. I was first given a new Solarwinds.APM.probes.dll file and instructions to replace the JobEngine35.sdf file with the blank version in the same directory. The immediate results were all my wmi performance counter…

The weird thing is that you say that CPU load on that system increases even when just 1 processes is polled from the box and CPU load stays that way. Are you sure this happens when you have just 1 SNMP Process Monitor set on that node? In that case it would look more like some issue in net-snmp because APM would poll the…

Is this still impossible? Would love to be able to customize the info displayed, custom properties in particular.

Is this still impossible? Would love to have this feature, not just for maps(hardly use maps..)

Update - the timeout issues appeared to have cleared up after a few hours. Typically on that server I'd get a half dozen or so polling timeouts for every component per day, so that's a nice improvement. However, now I'm noticing that all components on that server have the "next poll time showing a date/time earlier than…

Well I made the change as instructed in the KB, but the results weren't so good. I tried it using BulkRow with Bulk Size at 10 for an hour and then switched it to BulkColumn with Bulk Size @ 10 for an hour just for grins. End result - net-snmpd still uses the same amount of cpu, but now the snmp process monitor polls for…

+1

 +1

I did have net-snmp at the latest version, but I rolled it back to an older version to do a like comparison with one of my other linux servers that isn't experiencing the problem. I'm getting ready to make the change mentioned in the kb you linked now, will post back with results. Thanks for the reply!

+1, this would be very nice to have!

Disabling components was a nice addition, but it still doesn't give you the option to set a window for the disable. Also, by disabling a component it is effectively out of sight out of mind as it essentially dissappears from all of the Orion pages. At least in an unmanaged or unknown state you can set up a resource to show…

opened case #290761, will post the results

Josh -- Thanks for the reply. I believe that adding the ability to filter on something in the process command line would solve the problem, at least I know in my case it would. --Brandon ^^ What he said! That would definitely help us out as well! Thanks for the response Josh

Yep, it's a Dell R710 2 x Intel(R) Xeon(R) CPU E5620 @ 2.40GHz Quad core 96gb RAM 2 x 146gb SAS 6gbps in raid 1 - OS 6 x 146gb SAS 6gbps in raid 10 -SQL Win 2008 R2 Ent 64 SQL 2008 64 We've got 2 Orion db instances running on the server and the performance stats have been pretty good actually. The NPM/APM instance gets a…

A little more info - I ran a comparison between the problem server and another one with the exact same OS, net-snmp version, but much lower system specs than the problem server problem server - 1 proc monitored - net-snmp process cpu usage > 6% constant other server monitoring 30 + procs - net-snmp process cpu usage < 1%

Thanks for your quick reply Lukas! Much appreciated. Kris

Thanks for the reply. Will it also allow individual components to be unmanaged?

Thanks for movin that for me! As it stands I'm just going to make a dummy app for every server and point people to that. Kinda not so slick, but it'll get the job done unless I can figure out how to hack it up on my own

I appreciate the device poller info, but the problem here is the effect Orion has on the net-snmp process when trying to poll for processes on an extremely busy server. I'm looking for a more direct way to poll the server for particular processes that won't cause as much overhead as I'm seeing now. Or a more efficient…

+1 Yes it can be done with custom html resources, but it would be easier if all options were available no matter what type of view you're using.

It's not the answer unfortunately Danielle. When I originally built that script monitor it pulled the script output and found the unique identifiers just fine. However, now if you go back into that app and try to add or remove anything and update the script output it just fails. I even tried not changing anything at all,…

+1 I've had several of those embarrassing moments myself. We've had intermittent issues with disk volumes just deciding to stop polling for some reason. Usually that's exactly when some application goes haywire and fills up the c: drive on a server with log files(why people are logging on the c: drive is another issue..).…

Would love to see this for application/component outages as well..

Ran into that problem on our HPUX 11.11 servers. Using net-snmp version 5.4 and modifying the conf file as suggested works. Make sure to disable HPUX's version of snmp if its running.

73 million rows.. It's a pretty beefy SQL server. :)

For catching when your UPS goes to battery power you can set up a UnDP using oid - 1.3.6.1.2.1.33.1.2.2 name - upsSecondsOnBattery If the returned value of the poller is greater than 0 you're on battery power. I use a few others for our UPS' as well - upsAlarmsPresent oid - 1.3.6.1.2.1.33.1.6.1 Obviously anything over 0 is…

It would be nice to see more granularity out of the box. We have several procs with the same name that run in different different environments or get started with different switches. All of which show up when displayed on the machine locally. We can't really rely on the pid since some of the procs get restarted frequently…

opened case #290761. Here's the full original script that actively polls and works just fine unless you try to manually get script output from within the edit app function(not really necessary, just tried it to see if it would work in case I wanted to add/remove something from script at a later time). #!/usr/bin/ksh #set…