Two servers, one pool - failover, not load-balanced?!

what about this?

Single Node Persistence

That looks good, but I was hoping for something a little more simple.....the fact an iRule is necessary suggests to me that there is no checkbox for this feature then?

i understand destination address persistence should do the same thing.

Perhaps I have it working anyway, please have a look at What I have configured right now for my pool:

Server Name     IP Address  Ratio   Prio Grp
------------------------------------------------
Server1:4080    192.168.50.1    1   5
Server2:4080    192.168.50.2    1   0

Priority Group Activation = Less than, 1 available member.

This should mean that Server1 is prime and remains prime until it goes down. At this point the LTM should flip to Server2. The reason I'm questioning the above is that when I look at the Pools statistics I see traffic to both Server1 and Server2!

the point is when server1 comes back up, existing connection will still go to server2 but new connection will go to server1. if this is acceptable, your configuration should be fine.

as using priority groups still partially shares load

That's not true at all. There should be no load balancing in priority group activation, unless multiple servers are in the same priority group. Only the servers in a higher priority group will be used until all (or some defined count of those) become unavailable. That said, priority group activation makes no sense without some form of active monitor. Is it monitor traffic that you're seeing?

OK thanks again. Traffic does seem higher than just monitor traffic on the secondary server. How can I get more verbose on the stats?

P.S. I'm guessing 'Load Balancing Method' is irrelevant in my scenario with only 1 server in each Prio Grp?

Can anyone comment on my last question (above) please? Thanks

How can I get more verbose on the stats?

you may try tcpdump when stats is increasing. if it is monitor traffic, source address should be non-floating self ip (i.e. not floating self ip, snat ip or client ip).

Thanks nitass :) ...and the final one was, "I'm guessing 'Load Balancing Method' is irrelevant in my scenario with only 1 server in each Prio Grp?"