Unexplained Plex Media Server hangs / long pauses

I have a couple embedded PMP’s in my house and a PMS running under a dedicated virtual Debian 8.5 server (Linux plex 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt25-2 (2016-04-08) x86_64 GNU/Linux). The PMP version is Plex 2.10.8 on PMP 1.2.2 on OpenELEC 8.0 and the PMS version is 1.3.3.3148.

I have been experiencing for some time intermittent random hangs in PMS when using these PMP players. There doesn’t seem to be a hard predictable pattern to it, and I couldn’t say what initially triggered this to happen. I was using PMP 1.0 for 6+ months without issues, but recently upgrades to the 1.2+ versions with a full reinstall. The intermittent hangs started happening after that, although I couldn’t say exactly when.

The hang occurs in PMS when using the PMP player, typically while navigating menus or starting / stopping playback. If you just let playback run, it seems to work all right. PMS will appear hung, even though the PMS Debian machine is running fine and you can SSH into it. The Plex web client will say it can’t reach the Plex server, and the PMP client will say the same thing. If you wait long enough (maybe 5+ minutes) the clients will start saying they can reach the server.

When PMS is hung, you can SSH into the box and the Plex processes appear to be running. You can do a service plexmediaserver restart command and there is a long 1-2+ minute delay while the processes restart. If PMS is not in this hung state, a service plexmediaserver restart command immediately restarts PMS.

I’m attaching logs from tonight when this problem just occurred. I looked through the logs and could not see anything obvious in them that would cause this problem. Some messages about “NetworkServiceBrowser: SSDP departed after not being seen for 20.970758 seconds” and some messages reading:
EventSource: Failure in ReadData (125 - Operation canceled). MyPlex: We appear to have lost Internet connectivity, resetting device URL cache. ERROR - EventSource: Retrying in 15 seconds.
I can’t say these are the cause of this problem, but they appear near time gaps in the log and look like some sort of network problem. There isn’t any dynamic networking going on for this box, so I have no idea what could cause them. Any help or tests I could try to debug this would be greatly appreciated!