I have a procurve 4104gl, which I have set up a LACP static trunk to the server. The the server has a Intel Pro 1000 NIC configured for static LACP. The connection is great until I reboot the server. After rebooting the switches "show lacp" command shows that the trunk to the server is blocked. I have no STP running, and it only happens after shuting down or rebooting the server. To fix the connection I have to disable and then re-enable the ports in the switch. Does anyone know what is happening?
>I have no STP running Run a network trace to be sure. More often than not, BPDUs are discovered to be floating around where none were thought to exist. Maybe you will even find some rogue WAPs. ;)
I sniffed my network using Ethereal. I saw that BPDU's are being sent across the network. I have verified all of my switches have no STP running. The sniffer gives the mac for an HP device, and I can't find it in any mac table on any of my switches. Any thoughts?
I still have a problem with the static lacp connection to the server. The NIC seems to disconnect for about 10 seconds at reboot. I was going to try another NIC to see if that will fix my problem. The wierd thing is if I unplug the connecton to the switch before rebooting, and then plug it back in after reboot, the connection comes up. It seems some kind of parameter is being passed at POST boot of the server.
Going back to what Carsten mentioned, have you tried just leaving the switch ports as their default of passive LACP?
When configuring NIC teaming on ProLiants in the past, I've just set them to 'Automatic' and left the switches as is. The ProLiant will go into an LACP Active mode and the switch should then create a dynamic LACP trunk. I've done this a fair few times in the past with different model ProCurves and it's worked great. I also tried with Intel ProSet which worked well.
