LTM Policy - LB does still use a policy rule with an forward to pool action when all pool members are down
Dear all,
I've a strange behavior with my LTM running on 11.5.1HF5.
The following polices are currently configured :
Strategy : First Match Requires : HTTP Controls : Forwarding
Rules deployed:
Rule 1 | http-uri path starts-with /test/production1 = Action forward select pool /Common/prodpool1
Rule 2 | http-uri path starts-with /test/production2 = Action forward select pool /Common/prodpool2
Rule 3 | http-uri path starts-with /test/production3 = Action forward select pool /Common/prodpool3
Rule 4 | http-uri path not starts-with /test/production1
Rule 4 | http-uri path not starts-with /test/production2
Rule 4 | http-uri path not starts-with /test/production3
Rule 4 | = Action forward select pool /Common/prodpool4
Every pool just consists of one pool member, because of https://support.f5.com/kb/en-us/solutions/public/14000/900/sol14968.html
So we have pools configured as follows :
prodpool1 = prodnode1 prodpool2 = prodnode2 prodpool3 = prodnode3 prodpool4 = prodnode4
All those 4 nodes are still within a common pool called prodpool_all which is bound to the VIP as default pool.
Now for the problem :
If for example prodpool4 goes down ( because prodnode4 is down ), the loadbalancer is still forwarding traffic to this node, as the counters for the default pool ( prodpool_all) are not raising up. Therefore we can still see that the "Invoked" and "Suceeded" counters for Rule4 are counting up.
From my understanding i would have thought that if a pool which is used within a policy rule, this rule will no longer be evaluated. Can someone light up the darkness here ? Do we miss something here ?
Btw we also tried creating a catch-all rule at last ( with forward action -> prodpool_all ) and changed the strategy vom First-Match to All and Best, but that doesn't helped a bit.
Thanks a lot for your help
Best Regards
Christopher