Keith_Richards_
Jul 05, 2007Nimbostratus
LB_FAILED event
Please can someone tell me how long the LTM will wait, given that the selected server hasn't responded, before triggering the LB_FAILED event.
I have noticed that when people are using LB_FAILED for passive monitoring they also put in an LB:detach and LB:down. I don't understand why you use the LB:detach - is it just good housekeeping on the LTM? When we used the LB:down on our system there were pool members getting removed due to slow response even though they weren't really down/offline.
Oh, and sorry, but another thing. I'm using universal persistence as below:
when HTTP_REQUEST {
set uri [HTTP::uri]
set jsess [findstr $uri "jsessionid" 11 "?"]
log local0. "Entering REQUEST, jsess is: $jsess"
if { $jsess != "" } {
persist uie $jsess
}
}
when HTTP_RESPONSE {
if { [findstr $uri "jsessionid"] > 0}{
set jsess1 [findstr $uri "jsessionid" 11 "?"]
log local0. "jsessionid found, jsess is: $jsess1"
persist add uie $jsess1
}
}
The virtual server uses a second iRule for LB_FAILED events. I have seen other examples in the forum delete the persistence using a 'persist delete' statement - should I use this? I assumed that if a connection to a member fails with event LB_FAILED that that connection will be reselected (as below) and a new persistence record created?
when LB_FAILED {
set selected_server [LB::server addr]
if {$selected_server == ""} {
log local0. "No mdex node available"
}
else {
log local0. "Node: ${selected_server} not responding."
Select another node
LB::reselect
}
}
Thanks, Keith