1)
A lost packet will always qualify as greater than whatever number you enter. So lets say you set up an alert to fire when 2 of the last 10 samples. If you get 2 high latency packets (say 800 and 900), the alert will fire. In addition, if you get 2 lost packets, the alert will fire (ie: 20% packet loss in this case).
If you use 9999 instead of 500 for the latency number, then you'll be looking for packet loss - not just high latency (as the default time to wait for a packet is 9999 ms - anything more than this is a lost packet).
So using this same mechanism, you can trigger on high latency - which always includes packet loss, or just packet loss. You can get to whatever percentage of target packet loss you want to by changing the number of samples to look at and to alert on - ie: 23 of the last 100 would be 23% packet loss, and 155 of the last 1000 is 15.5% packet loss (although that's an unreasonably high number of packets to look at).
2)
We're currently toying with some ways to change this a bit, but haven't come up with a perfect mechanism yet. Are you suggesting that we just add a menu option (from the main menu) for alert association with the currently selected host?
Thanks for the feedback.
- Pete