I guess that the problem is this: a pool is responsive to pings, but however (for any reason) it cannot offer you any mining work, so miner is idle. And your program decides to keep trying to mine here and not making an attempt to switch pools. Am I right on this? 
Yeah, still can't figure out a better way of checking if pool is down or not. So, it's either ping or checking for share acceptance rate, unless someone comes up with a better idea, preferable one that doesn't need separate implementations for different pools.I vote share acceptance rate, after all that's the true measure of the value of the mining process.
So, if it's share acceptance rate and not network reachability, how do you decide when to switch back to the primary pool? Perhaps in that scenario of handling failover, you would be able to set a time until the primary is tried again (similar to the current setting), and a configurable period of time during which the primary must accept a share or it goes back to the secondary.
You know what I'd really really love more than anything else for this program? A simple button to force a client to switch from primary to backup or backup to primary. Right now, I do it by swapping ini files.
The other thing I would love would be if the program remembered where your OSDs live. Having to reposition them every time is frustrating.