Connection limit logic

This may sound like a very simple question, but a lot of googling has not yet given me an answer, so I'm hoping someone here knows.

I'm investigating the logic of node connection limits when the node is a member of multiple pools.

We have about 20 different websites all being served by 5 servers using host headers. Each has its own virtual server and pool for statistics collection purposes.

The member connection limits are set to zero on all pools, LB method is Observed (Node) on all of the virtual servers.

Each of the 5 nodes has a connection limit of 250. No connection limit is set on members within the pools.

My expectation (hope) is that it's going to be first come first serve... when a node hits 250 open connections, regardless of the fact that those connections are from several different pools, that node will become unavailable and refuse new connections from any of the pools until the number of open connections drops below 250.

Should it be working as I expect?

There is a reason I ask... after being up for several weeks, I looked at my statistics today and noticed that all of my nodes have a maximum connection number well higher than 250, in one case as high as 419.

Thanks!