Forum Discussion
jonnyf5
Nimbostratus
Nimbostratusuneven log balancing and elevated processing time after node reboot
facing an interesting issue version - BIG-IP 14.1.4.6 Build 0.0.8 Point Release 6 iapp > pool > i7 rule > if traffic matches i7 uri rules then it's redirect to "subpool" "subpool" consists of 2 n...
JRahm
Admin
AdminHI jonnyf5, the 120 minute persistence is very likely the culprit. When one of two servers goes down, the remaining one will take all connections while the other is out of service, and retain those connections once it's back as long as the clients are hitting that virtual within the 2hr persist timeout (which is not absolute, it resets on each connection).
I'd need to know more of the why of your scenario before making recommendations, but off the top of my head, you can likely use an iRule to track the two server LB::status state and then when there is a transition from false->true, you can issue a persist none on the one that was active solo.
JRahmAdmin
AdminAnother approach you can explore is using the PERSIST_DOWN event to add any clients making a request to the server that's down that had a persistence entry to a table, and then you can on future requests do a table lookup against the client IP and if present, AND the server is back up, send them over.