Server Version#: Linux-1.16.5.1554 (remote, updated recently) and MacOS-1.9.7.4460 (local, not updated for a while)
Player Version#: AppleTV 2.7
Hoping to get some tips or at least corroboration here. I have two PMS instances: One is local (a MacMini, running 1.9.7.4460, physically in the house, on the same network as the AppleTV) and one is remote (on a Linux VPS, in the cloud, running 1.16.5.1554). Operation has been mostly rock solid for a long time (years) with the local server, but has been extremely problematic with the remote server. Here are some of the issues I’ve seen with the remote server, but not with the local server:
- Thumbnails are unreliable. Many thumbnails appear missing in the client.
- Metadata is generally much more sparse on the cloud server (i.e. episode descriptions are often, but not always, missing)
- Trying to use the client while the remote server is re-scanning the library is effectively hopeless. This use case has always been a little troublesome, even with the local server, but with the remote server, the ATV client app will sometimes become unresponsive, and I’ll have to kill and relaunch it.
- “Watched” status is almost never correctly set. I’ll watch a full video, but it will show progress as 5 or 10 minutes in. I almost, but not always, have to go in and manually mark things as watched.
- After manually merging two shows in the web interface, they become un-merged upon the next library scan. It’s not just merging either; manual metadata edits will disappear too.
- The server seems to re-scan the library a LOT, and the scans seem to take a long time compared to my local server (despite being a much smaller library and a much faster computer)
Things I’ve looked into:
- The bandwidth between my network and the remote server is fine. Once I’ve actually started playback, I’ve never had a problem. I can reliably saturate my home internet connection when talking to the VPS. (Not when using Plex, but, say, doing an
scpbetween here and there.) - I’m suspicious that latency might be an issue, and that the ATV client is giving up on requests very quickly, but I have no easy way to test this.
- Re: metadata, I’ve tried putting the same media file on both servers and the local/MacOS server does get good metadata, and the remote/Linux server, in some cases, does not.
- I’ve turned the periodic re-scans down to daily – we’ll see if that helps.
Other observations:
- The client-side usability issues seem to be specific to the ATV app. I’m able to use the web interface and the iPhone app more reliably (although the web app is also less usable than you’d like during a library scan.)
- I can’t pinpoint exactly when, but I feel like a while back the ATV app was much more reliable, then I updated it, and things started getting wonky. The thing I remember the most is that the “Reload” button disappeared, and we frequently used it. Now, to “Reload” we kill the app and restart it - takes a lot longer, and is more cumbersome than the Reload button was.
- Being a software developer by trade, I have a strong suspicion that there’s some sort of caching problem in the ATV app. Of course, cache invalidation shares the throne with “naming things” as one of the greatest problems in computer science, so…
Anyway, I realize that there’s not any technical data here to go on in terms of a concrete diagnosis, but I wanted to put this description out there and see if other people were having similar issues. I’ve really enjoyed being a daily Plex user for years, and I thought putting a server in the cloud on a VPS would solve some logistical problems for me, but it seems to have brought quite a few new issues with it.
Thanks in advance for any thoughts/tips.
