How do I create a passive health monitor?

Actually, there is a new command in 9.4.x called LB::down. Basically, when it is called, it forces LTM to mark the node down and schedule an immediate health check using the assigned monitor. LB::down does NOT force re-loadbalancing itself - if you want to repeat your request and force another LB decision after calling LB::down, you can issue HTTP::retry.

 

Here is an example:

 

 

when HTTP_REQUEST {

 

set request [HTTP::request]

 

}

 

when HTTP_RESPONSE {

 

if { [HTTP::status] >= 500 } {

 

LB::down

 

HTTP::retry $request

 

}

 

}

 

 

Any updates from the support request vaniello? Over a year later and there is some misleading information here from my attempts. I'm new to this and perhaps don't know the method.

 

 

Findings:

 

 

If the health monitor still thinks the member is available, it seems things aren't going to go well.

 

Is there a way to makr the node as down faster than waiting for the health monitor cycle to do it?

 

 

the LB::down marks it down, true, but it isn't a health monitor down, it is a irule down. So then the question is how does that member become available again?

 

 

LB::down doesn't seem to keep the member from being reselected with a LB::reselect if there is 'cookie insert' (or probably other) persistence set. It will continue to reselect this same, failed member.

 

 

LB::detach does not appear to clear the persistence either. using that with a LB::reselect results in selecting the same member. the LB::detach doesn't help.

 

 

Looking for a code sample where another member is selected and the persistent cookie is updated to this new member if the 'assigned' member is no longer available.