Have a wacky problem and I'm hoping someone out there can shed some light on it before I start babbling to MA support about it. So first: some background info:

Problem: Session between a single MA2 Light and an NPU (both running 3.9.60.27) is proving unstable and slow. First noticed it 2 days ago when the console briefly lost control of the rig. We thought we had resolved the issue with several hard resets. However, during show in afternoon we experienced several minutes of what appeared to be excessive panicked network traffic on the NPU. Among the traffic I noticed a lot of "Nack recieved" and "jobque overflow" messages at the rate of roughly 60 per second. Also of note were several Session Lost/Session Restored messages that were milliseconds apart from eachother (so fast that the heartbeat was never noticeably lost at the console and NPU. At this point the session was still functioning, but obviously unstable.

Troubleshooting action taken:
1.Soft Reset the MA2 Light at FOH.
-When it rebooted, console immediately bricked when it attempted to restart the session.
-NPU stopped rapid firing error messages at this point

2. Hard reset MA2 Light and fiber rack at FOH
-When everything rebooted, the session restarted and appeared to be a bit more stable.
-NPU downloaded the show verrrrry slowly and began repeating its same "NACK" error messages, but at a noticeably slower rate than before.

3. Hard reset NPU and fiber rack at distros
-When everything rebooted the NPU joined session and downloaded the show at normal rate.
-At this point the NPU reconfigured its outputs to default settings for some reason.
-Once reconfigured to the proper outputs, the session became stable and we have had no probems since.

Has anybody else experienced this? I'm only putting this post out there at all because this is the 2nd time in 2 months that I've had similar session issues while using MA2 consoles, NPUs, and fiber switches. Not sure what is causing thi, but after consulting a couple of wizards offline I'm leaning towards the managed fiber switch being the root of the problem.