I recently tested this balancers (two – cause redundancy). I liked these devices, especially for iRules functionality. But I strongly disappointed about his realisation of redundancy (Active-Standby). If we want to determine what device from redundancy couple should be active, after reboot this “active” device redundancy scheme don’t work correctly.
From tests with traffic dump that we can see:
1. Start step. Every services works fine, device “A” is active, device “S” is standby.
2. Device “A” is rebooting. .
3. Device “S” after fact that “А” is down, becomes active and send gratuitous arp for every virtual ip..After one second all services are up and running correctly.
4. Device “А” is returning to network after reboot and want to be “active” again. Device “S” throws off his duties (no longer serve services), but device “A” dont send gratuitous arp, therefore all virtual ips live with old mac in the arp-cache and every service don’t work.
This situation can be fixed only if you determine for all device redundancy as “auto”. After that device was “active” after reload don’t want be “active” again and stays “standby”.
This problem is very serious cause it breaks idea of definition what device should be “active” ny default. And I don`t understand why this function exists in the options after that.
This problem was observed on the BIG-IP PSM version 10.2.3