Active / Passive slightly diffrent recovery

That's sounds like a standard HA deployment. No, iRule needed. Both devices in an HA pair do not make connection to the pool members. Even if connections are mirrored, its just mirroring the state table not created a second connection.

That's sounds like a standard HA deployment. No, iRule needed. Both devices in an HA pair do not make connection to the pool members. Even if connections are mirrored, its just mirroring the state table not created a second connection.

Ahh, makes much more sense now. Probably the best/easiest way to accomplish this would be to use priority activation and manual resume on your monitor, but that still would require manual intervention to fail back to the primary pool member. I haven't tested this yet, but give it a try if you have somewhere you can test it. Uses tables and swaps back and forth when LB fails.

when RULE_INIT {
    set static::poolA "/Common/poolA"
    set static::poolB "Common/poolB"
    table set primaryPool $static::poolA indefinite
    table set secondaryPool $static:poolB indefinite
}

when HTTP_REQUEST {
    pool [table lookup -notouch primaryPool]
}

when LB_FAILED {
    set origPrimary [table lookup -notouch primaryPool]
    table replace primaryPool [table lookup -notouch secondaryPool] indefinite
    table replace secondaryPool $origPrimary indefinite
    pool [table lookup -notouch primaryPool]
}

Ahh, makes much more sense now. Probably the best/easiest way to accomplish this would be to use priority activation and manual resume on your monitor, but that still would require manual intervention to fail back to the primary pool member. I haven't tested this yet, but give it a try if you have somewhere you can test it. Uses tables and swaps back and forth when LB fails.

when RULE_INIT {
    set static::poolA "/Common/poolA"
    set static::poolB "Common/poolB"
    table set primaryPool $static::poolA indefinite
    table set secondaryPool $static:poolB indefinite
}

when HTTP_REQUEST {
    pool [table lookup -notouch primaryPool]
}

when LB_FAILED {
    set origPrimary [table lookup -notouch primaryPool]
    table replace primaryPool [table lookup -notouch secondaryPool] indefinite
    table replace secondaryPool $origPrimary indefinite
    pool [table lookup -notouch primaryPool]
}

Some experimenting later. I'm sure there is room for improvement, but it seems to do the job.

when CLIENT_ACCEPTED {
 Active/Passive iRule. IJdo Dijkstra 15-10-2015
 Pool A should be active, and failover to B. Pool B should remain active even if pool A returns.
 Failback occurs when pool B is down. 
 User table entry FailOver for failover state. 0 = normal, 1 = Failover
 Checked for FailOver!=1 to allow for non-set condition.


 Normal situation. Pool A is up, and FailOver != 1. Traffic to pool A
if { [table lookup -notouch FailOver]!=1  && [active_members /Test/pool-test-A]!=0  } {
    pool /Test/pool-test-A

 Normal failed-over situation. FailOver = 1 and pool B is up.
}elseif { [table lookup -notouch FailOver]==1  && [active_members /Test/pool-test-B]!=0  } {
    pool /Test/pool-test-B

 Actual failover trigger. Pool A is down, FailOver != 1 Traffic to Pool B
} elseif { [table lookup -notouch FailOver]!=1 && [active_members /Test/pool-test-A]==0 } {
    table set FailOver 1 indefinite
    pool /Test/pool-test-B

 Failback when pool B goes down and FailOver is 1. Traffic to pool A
 No strict need to check if pool A is up (with both pools down service is lost); but it seems neater.
} elseif { [table lookup -notouch FailOver]==1 && [active_members /Test/pool-test-A]!=0 && [active_members /Test/pool-test-B]== 0 } {
    table set FailOver 0 indefinite
    pool /Test/pool-test-A

 Happens is all the above conditions are not met. Optionally additional conditions could be used for sorryservers.
} else {
    pool /Test/pool-test-B
} 
}