Forum Discussion

theanswriz42's avatar
Jan 26, 2024

F5 Virtual Edition in Azure - Strange Behavior with Sticky Sessions on IIS App

My organization currently has a stateful application running in Azure behind a F5 VE deployment configured for sticky sessions and round-robin. We're seeing strange behavior where a node is marked down in the pool (ie. service is down on the host), but we believe the LB is still trying to send traffic to the downed node due to the cookie not getting invalidated (totally my assumption), and I was wondering if there's a specific config somewhere in the appliance to force a new cookie to be generated in the payload to send traffic to the available back-end node in this type of event? 

 

Also, if there's any documentation covering these types of configs and scenarios, I'd be very appreciative. 

 

Thanks in advance! 

1 Reply

  • This could be related to the 'Action on Service Down Feature' which is set by default to none (which essentially means existing connections are still sent to the down pool member until the connection times out, but new connections are sent to a different pool member).

    I would try experimenting by changing the action on service down setting on the pool to reject (reset) and see if it makes any difference.

    https://my.f5.com/manage/s/article/K15095