LACP Port-channel int in use when partner server not speaking LACP - is this normal?

Hi all,

I have a 2 x PC5448 stack, and have configured 6 port-channel interfaces (2 phy int's each) with the command 'channel-group {#} mode auto' on each phys port. When I then do a 'sh int status' I can see the instantiated "Po#" ports at the end, in 'down' mode. Of course, when I configure the attached server (RedHat Linux OS) to bond two of its NICs and use LACP for the proto, it works just fine. However, when I put a server on just one of the two phy int's that were configured with the LACP auto command, and the server is NOT using LACP, the "Po#" interface is still the one in the forwarding table when I do a 'show spanning-tree active' command. I am used to Cisco switches that only form the "Po#" interface when the partner system negotiates the channel, and they just use the phys int(s) in the forwarding table if LACP negotiation doesn't happen. It doesn't seem like the use of the "Po#" int rather than the single "Gi1/0/#" phys int is causing a problem, but it's just unexpected. Is this the way it's supposed to work? 

Thanks,

Will