Unmanged Nodes still being polled?

Question

Now that I have Netflow 2.0 running on my network, I have been noticing something that kinda bothers me.   In order to keep a history of nodes retired and or replaced, I have been unmanageing those nodes and renaming them starting with a Z so that they fall to the bottom.

Now after watching Netflow, I am seeing SNMP traffic going from my Orion NPM poller to those nodes.  This makes me think that the pollers are ignoring the unmanaged flag I am setting.  Is anyone else seeing this?  Is this a known issue or something that is only affecting me?

Thanks,
MarkS

jtimes · Answer

I also see polling occasionally going out to unmanaged nodes.  I only unmanaged nodes that are due to be repaired after several days or weeks, depending on the issue.  If a node has been permanently removed from the network it gets deleted from Orion. 
I honestly don't think that Orion was ever intended to be used as an inventory management solution. Cirrus is more suited for that purpose.

John J. Times

salyerma · Answer

I am now convinced that the poller does ignore the unmanaged setting.  We decided to delete all of our old unmanaged nodes and things improved greatly.  I had somewhere close to 200 unmanaged nodes that I deleted.  The website showed us monitoring less nodes than before.  Plus, the poller tuner showed a lot less nodes and interfaces (-5000) and was able to complete a round of polls in time to not have gaps in our graphs.   We always had ~500 SNMP statistics polls outstanding.  Now we are down to only ~100 which is close to what we poll.

Also the tuning wizard asked for a lot less polls per second than it did before.   If you have a lot of unmanaged nodes, you may want to keep this in mind.

Now a question.  I have been using unmanaged nodes in order to keep a history of testing and/or status for auditing/SOX purposes.  Are there any plans to improve the unmanaged piece so that those devices are ignored when it comes to polling?  It really made a huge difference to just delete those nodes, but that is not the best answer in my situation.

Thanks,
MarkS