I was sent the following by my boss:
a couple weeks ago, i got around 50 emails because the t-1 between internap and atl was down. every server in the atlanta office generated an alert because the nimbus server couldnt communicate with them. if the t-1 is down, we should get an alert saying the t-1 is down. the other machines arent down, its just that the nimbus machine cant see them because the t-1 is down. so in this case, set up a dependency that says the machines in the atlanta office depend on the t-1 being up. if the t-1 isnt up, dont send an alert. if you have documentation from the nimbus class, im sure it goes over that...pretty standard for monitoring schemes.
Does anyone have any suggestions on how to implement that? Thanks!