We are working with a Sun M5000 (connected to a Cisco 6509) server and are utilizing 2 gigabit cards to handle 2 lacp aggregates as follows
Key 1: Public routed ip address Policy: L4 LACP Mode: passive lacp timer: Short e1000g0 e1000g2
Key 2: Private ip address only for Oracle RAC heartbeat Policy: L4 LACP Mode: passive lacp timer: Short e1000g1 e1000g3
On our aggr that is the public ip, every so often we lose one of the port channels, the output from dladm show-aggr shows one member of the pair has gone to
Link - unkown State - standby
At this point our bundle is only running on one leg. The network team then will bounce the ports on the switch and this brings it back into a bundled "normal" state.
We have 2 Sun M5000 in the same configuration and it happens on both. We never lose the channel on the private Key, only the public ip one.
We have done a lot of troubleshooting of moving wires, on the server and the switch side, to see if this helps, all to no avail. We have even gone so far to move both channels of the aggregate to the same swicth in hopes that we were having an issue with VSS on the cisco side. but it failed again after 8 days
Any thoughts or suggestions on troubleshooting or ideas are greatly welcome as we have seemed to hit a dead end