Server Version #: Anything over 1.19 (Currently running 1.20.0.3181)
Player Version#: Any Remote Access
I wanted to give a head’s up on PMS 1.20.0.3181 for QNAP NAS running 4.4.3.1400 (TVS-873 AMD RX-421ND 64G and TVS-871 i7-16G) for remote access. Any remote connection causes the CPU to peg 100% CPU usage and an eventual NAS crash. This started happening as soon as I upgraded from PMS 1.19.
For background, I have a Lifetime Plex Pass and have been running PMS for a few years over several versions. Some of my friends and family download/sync to their devices while others stream from my server. Just since the upgrade from 1.19, the ones that download/sync are being transcoded to the max streaming quality (they used to be able to get full quality) and the streaming connections have gone from playing nicely with transcoding several simultaneous sessions to using the entire server CPU resources in just one stream. Even a single remote streaming session will cause the NAS to crash after several minutes of transcoding. In house I use direct play and have no issues.
I have been able to work around the issue by running a docked version of PMS (1.19.1.2645) to provide the remote access until the latest version is fixed, but since I haven’t seen a lot of others reporting this issue, I wanted to be sure that it was documented so others would be aware before upgrading. Once the upgrade is done, installing the older version over it does not fix the problem.
The docked version of Plex (1.19.1.2645) running on the TVS-871, has the the virtual CPU set to 50% with 8G RAM. It runs great, remote streams play at the maximum streaming quality and download/sync does not transcode. Version 1.20.0.3181 running as an app on the TVS-873 NAS consistently crashes when transcoding and all download/sync are transcoded.
Should I expect that any new versions of PMS will transcode all download/syncs? I would like to be able to turn that off or at least allow the selection of full quality, non-transcoded for the friends and family that want it.
Is the 100% CPU utilization due to changes in the transcoding process? I would ask that the developers review any changes to confirm that there are runaway processes with the newer version and fix them.
Head’s Up for those who are running PMS, and Thank You to the developers who support us!
Without knowing the particulars of the actual media being streamed and what device(s) being streamed to, it’s not possible to know what should be happening.
Hi Chuck,
Thanks for the reply. To clarify, I have two QNAP NAS:
TVS-873 running QOS version 4.4.3.1400. Plex version 1.20.0.3181 was recently upgraded from 1.19.something. The application is manually installed and running on bare metal.
TVS-871 running QOS version 4.4.3.1400. It is running PMS 1.19.1.2645 as a Docker in Container Station version 2.1.3.1360. The Docked version of PMS is configured for 50% CPU and 50% of system RAM.
I set up system 2 because any remote connection to system 1 would consistently crash after upgrading PMS to version 1.20.0.3181. Downgrading to an earlier version does not fix the issue.
I reinstalled the latest version on system 1 and am only using it for local network direct play with OpenPHT, Windows client, and Web. All play perfectly. I turned off remote access as any connection will cause the CPU to run up to 100% utilization and eventually crash. Recovery from the crash requires resync of the raid volume and file system check and repair. Remote connections were made with the Roku client, Playstation client, Xbox client, Android client, Samsung TV client, Amazon Fire Stick client, IOS client and library shares to multiple other PMS instances. A check of running processes with remote access shows transcoding is tying up the CPU.
My initial test was to set up the PlexPass Docker on system 2. It uses the latest version of PMS and auto upgrades to the latest on every restart of the Docker. The initial version was 1.20.0.3181. Local direct play works fine. When remote access is enabled and a connection from any client or PMS share is made it will use every bit of virtual CPU. The video buffers to the remote system and becomes unwatchable. The docked PMS will eventually fail and become unresponsive.
The next test was to set up the 1.19.1.2645 Docker. It works perfectly for any local or remote client. Transcoding initially runs the CPU up to 70-80% utilization of the virtual CPU and then tapers back down to 10-20% as the stream plays.
Transcoding for streaming remote connections is the first and most critical issue.
Another noted issue is that download/sync to other PMS systems no longer transfer at original quality in PMS 1.20.0.3181. They are transcoded to whatever is set in the ‘Limit Remote Stream Bitrate’ setting. Version 1.19.1.2645 does allow original quality download/sync.
I have attached a log of playback from system 1 streaming to the Plex client for IOS. As expected, the CPU immediately shot to high utilization and did not back off until I stopped playback. I can do the same capture on system 2 if needed.
I’ve not seen or heard any updates on this topic. Does anyone know if the transcoding programming has changed significantly in any of the recent PMS versions that would cause forced transcoding on download/sync operations and increased CPU utilization on streaming transcodes?
Update for the notes here, The PMS host running on bare metal has been following plex updates and is now on 1.20.3.3401. The virtual PMS is still running 1.19.1.2645. I have tested transcoding and download/sync on each PMS host upgrade and have encountered the same issue.
Here is the cut and paste of what was in the ‘LAN Networks’ setting 10.3.13.0/24,10.0.3.0/24 no typo
The 10.0.3.0/24 setting was added after the issue started, to allow the virtual plex server to talk to the existing bare metal plex server. Just to eliminate confusion, both plex servers are now hosting in the same IP address space, 10.3.13.0/24. I think that is a red herring, but I will test and repost my logs.
