Is there a way to limit the number of pool members from being forced down by a health monitor?

It's not possible to implement this kind of logic with default monitors. You could do something like this in an external monitor which calls a shell script. You could use bigpipe commands (or tmsh in v10) to get a count of the active and total number of pool members and then call curl or nc to send the request to the pool member. But would you would need to be careful about when to not mark a pool member down when it's not responding. You could end up trying to send requests to a truly dead server.

 

 

Aaron