Forum Discussion
5 Replies
Sort By
- hooleylistCirrostratusYou could shorten the timeout to lower the time that a pool member has to respond to a monitor. You could also use an iRule to disable a pool member when a load balancing failure occurs using LB::down in the LB_FAILED event. Or you could use passive monitors. This post has some very useful info from Matt on passive monitors:
- Very cool, thanks, can I see an example of the iRule in action?
- hooleylistCirrostratusHere are a few related examples:
when HTTP_RESPONSE { Check if response code is a 503 if { [HTTP::status] == 503 } { Mark current server down LB::down } }
when LB_FAILED { Mark current server down LB::down }
- Do I have to key on the response code? That wont work for me if it's a different kind of monitor.. I should be able to key on a pool member status up/down correct?
- L4L7_53191NimbostratusOne note: I believe that when LB::down is called it'll trigger an immediate service check on the pool member, which can cause unexpected results if you don't keep this in mind. I'd definitely confirm this in your testing.