Hi all,
We're seeing an issue where after a reload or power cycle of our SG350's spanning-tree puts the trunk link into a broken/port inconsistent state. The only way to recover is a physical re-seat of the fiber, a shut/no shut of the port doesn't work on either end. Wondering if anyone has seen this issue before, it seems to be widespread, the native VLAN matches on both ends of the trunk, the spanning-tree state on the Cat9500 looks good, Control Plane and Data plane show operational and forwarding for the correct VLAN's. Looking at debugs it looks like the SG350 doesn't respond to the Rapid-PVST negotiation from the Cat9500, both switches are configured with rapid-pvst. One thing to note is that I can't recreate the problem outside of a reload/power cycle, so I can't capture debug on the SG350 side. The one thing I noticed was the SG350 defaulting to long cost rather than short, we've set that to match, but still no luck.
What's interesting is the port is up/up when this happens, the optic light levels look good, we've swapped SFP's and the issue still persists, we've moved ports and the issue persists. We've seen the behavior on multiple SG350's, so it seems like an issue between the SG350 and Catalyst 9500 specifically. We're currently running 2.5.7.85.bin on the SG350's, I see there is a newer FW version, but don't see any bug fixes that would be related.
Any help would be appreciated.
