Dynamic IP and releasing polling IP address?

I have a very similar problem - except - in our environment we are using DHCP (of the Windows Server variety).

Basically, the issue is as described above:

• a machine drops off the network (for one reason or another)
• Eventually, the DHCP lease expires and the DNS record is deleted
• A different machine grabs that IP
• SolarWinds now says the node is up even though it is definitely not the node that's responding 

When in this state: on the SolarWinds server, both ping and nslookup return no results for the old DNS name.  I can even display the DNS cache on the SW server and it clearly says "Name does not exist." for the old DNS name.

Yet SolarWinds still reports the node as being Up.  Since the node is set to Dynamic and is based on the DNS name, one would think it should be marked as Down at this point...

Help?

I think that's a very unique edge-case mgaciarz. In some ways it may make sense that if dns lookups fail it consider the node down, but may be extreme to generally do that as dns can fail. I have no idea if SolarWinds tracks these failed lookups though with an event or otherwise. If it changes, that's an event but not sure on a failed lookup. I recently built a decommission process into our SolarWinds, and I'm thinking maybe a similar type of thing could be done for your scenario. The alert could trigger if a node has been down for X amount of time.. for example let's say 10 hours -- something that may be just shy of the dhcp lease expiration interval. The alert would probably need to contain some interesting logic, and some level of an external program (powershell script) that does a manual query of the fqdn and/or looks at the polling engine dns cache as well and then set it unmanaged. Then send a friendly email to the email contact of the node etc.

Thanks for the idea.

I worked with support on this and they confirmed that the system is currently working "as designed" and it seems unlikely this will change 

The development team has created an internal issue for this tracked as DC-3436. By design, solarwinds performs the following:

1. Orion always uses IP address for polling and the one used for each poll is selected with following rules
   1. Static one for non dynamic nodes, i.e. the one stored in Orion.Nodes property IPAddress
   2. The one returned from DNS at the time of poll for dynamic nodes or previously used one when DNS does not resolve the hostname to any IP address. (Since the IP stays 'pollable' and passes WBEMTEST, it keeps the same polling IP and doesn't re-detect new IP)

If node is polled as dynamic, then first poller that is being run detects new IP address to use for the poll in all pollers

---

I've set up a SQL query to at least show me when I'm having this issue but may look at something more similar to your decomm process if/when I get some time.