Directly after a restart of the SpectroSERVER, about 220 invalid RTM Test Host models were created without a name. There are several of such models created for the same device. For example the VNM model got 41 of such "zombies".
Spectrum 10.2.x and 10.3.0
When there is a change in device IP address, than that IP is updated to RTM test Hosts, but not to the RTTMON application model. So on SpectroSERVER restart and on finding the RTM Test hosts with the application model, it uses the old application model's IP address, to find for existing RTM test hosts. Since there will be no test hosts with old IP address, it creates a new one always on restart of the SpectroSERVER, along with the existing RTM Test Hosts.
This issue has been addressed in Spectrum 10.3.0 in the 10.03.00.BMP_10.3.001 or higher 10.3.0 BMP or Spectrum version.
As a workaround, you can do one of the following:
Update the application model IP address when there is a change in the device IP address. so that model handle of updated application model will be used to identify existing RTM Test Hosts. Already existing duplicate Test hosts need to be "deleted" after setting their attribute "test_app_mh" to 0. Then this fix can be applied to avoid duplicates in future.
Destroyed all duplicate IP models after the next SpectroSERVER start. Spectrum will create a new RTM_TestHost model.
If in the event they do return such as noted in this KB, and setting the test_app_model to "0x0' still does not solve, then you may have to delete the RTTMON app instead.
Do a Locator Search for Application > Name and search for RTTMON.
Filter the results on Device Name - any that show with a "blank" or empty Device Name is a duplicate/bad model. Delete these and restart SS.
This was seen on 10.4.2 which should have been resolved. but may be a separate issue this time.