PR: https://github.com/OpenNMS/opennms/pull/1954

As part of this, also updated Response Status code to 'Accepted' (202) for delete calls which are asynchronous.

In fact, it seems like the new behavior is the opposite. You can only delete monitored services, and if after that the IP interface has no services, the IP is automatically removed. Then, if after that the node doesn't have IP interfaces, the node is automatically removed (I just tested this using ReST).

What I've said about requisitioned nodes is still true.

That's interesting!

I recall that on older versions, deleting the interface also deleted the monitored services inside of it, which is not happening at the moment, but at least there is a workaround.

Of course, if we're talking about an IP that is explicitly part of a requisition, I'd expect that re-importing the requisition will re-create the IP. That being said, for requisitioned nodes, if the IP was explicitly defined on the requisition, it should also be removed from the requisition.

I was having this issue on multiple versions of Horizon, but found something interesting today.  If I deleted all the services from the IP, the IP interface would delete off the node.

I would have expected the behavior to be if I delete an IP, the associated services would automatically be deleted.

On a VM with latest Meridian 2017, I found that regardless if we're talking about an IP interfaces that is part of a requisition, or an IP discovered through SNMP, it is not possible to delete IP interfaces through ReST or the WebUI.