Usually I would create a Pingable model with an IP address to monitor devices I'm getting traps for but can't actually monitor.
The problem is we don't have access to the IP address and are using Spectrum to monitor devices through an EventAdmin model.
Is there another model we can use that won't try to ping the device? We don't want Initial models and are hoping that there is a 'special' container type that uses the IP but doesn't ping it.
Is there a way to do this without pingable initial models?
The only model I can think to use is the EventAdmin model. It has an ip address for trap processing but does not ping the ip address.
Yes we have to use the EventAdmin models for the initial processing, but if we want alarms under specific models we need to have some model which has an IP to 'forward' the new alarm to.
The Pingable model is what I've used before but having so many Initial alarms isn't idea. I'm also sure it affects the startup times of the landscape a lot as it is trying to ping these devices and timing out a lot (and retrying a lot!).
You have EventAdmin model and EventModel. If the device is not modeled in Spectrum and one alert is created for that specific unique identifier (either IP address or another way of identifying uniquely each specific device), then Spectrum creates a new EventModel and asserts the alert/alarm on it.
Isn't this the default behaviour for EventAdmin models?
Correct. You are probably using Event Variable 8 to have South Bound Gateway forward the alarm to a model with an ip address. You can use Event Variable 1-6 instead to have a unique Event Model created inside the EventAdmin model instead.
I'll have a look. I think Event Models is what I'm looking for - thanks for the advice.
There's a case study in the documentation. I did an integration like this for a ManageEngine AppManager integration.
In all my years I've always been able to ping things so I think this is what threw me off. Joseph hit the nail on the head with varbind 8 My standard MO!