Mon Feb 28, 2011 5:56 pm
I was suggested a while ago to use a script that increases or decreases the max-limit of a queue based on ping-latency.
For start, I've tried to implement some scripts that activates a queue whenever Traffic Monitor is over 0bytes for a specific interface, and disable it with a 2min delay when it was 0bytes.
1) traffic >0byte on specific eth, run script that activates Queue
2) traffic =0byte on specific eth run script that waits 2 mins and then it disables Queue
IMHO, the problem was that Traffic Monitor keeps track of all the past past events, and combined with that script that has 2 minute delay (sleep(120)) before it disables Queue, I'm getting enabled/disabled queues from past events, instead of actual events.
If I wasn't clear enough, I'll describe a simple scenario:
There is traffic on eth, meaning that the Queue should be activated, but the Queue gets disabled because 2 mins ago there wasn't traffic.