Forum Discussion
3 Replies
- Stanislas_Piro2Cumulonimbus
The irule is doing persistence based on a static record.. 1
The first connection will select pool member and create a persistence record... every other connections will be redirecting to the same pool member until persistence record timeout.
- cmard_195831Nimbostratus
Thank you Stanislas,
My next question is which parameters in F5 (either from the command line or GUI affect this persistance record timeout.
In our case we have 2 x tomcat servers and only one is the one "in persistance" no load balancing. When the F5 detects that the tomcat is down (port 9090 for us), the F5 starts sending packets to the other tomcat, which was in stand-by mode. The fact is that there is a period of time (30-40 seconds) that the F5 is sending packets to BOTH tomcats and thus there is a mess. The fact is that the F5 does not seem to make a "sudden death" situation where if A is active, I send packets only to A and when there is a switch over then packets are sent only to B. There is a gray period on this. I do not know by how much the above irule is helping.
Awaiting for any suggestions. Many thanks
- Stanislas_Piro2Cumulonimbus
I did not find the default timeout. it may be 30 seconds as you saw.
you can force timeout time with
. 300 is the timeout in seconds.when CLIENT_ACCEPTED { persist uie 1 300 }
if you want to configure HA, the solution is to use priority group activation in pool configuration.