Hi all,
I’m running the most recent version of plex on Ubuntu cosmic so far. My media folders as well as the recording targets of the DVR are located on a NAS connected via smb and 10gbps. Everything has worked nicely so far. Due to increased storage consumption for previews a.s.o. I decided to follow this guide (Moving PMS 'Library') and move my Library folder from /var/lib/plexmediaserver/Library to the NAS as well. Problem now is: Everything is working fine except if I record something via DVR the *.ts-file shows up on the NAS correctly and can be played i.e. via VLC, it’s also showing up in plex overview but when I try to start playing it via web or android app it gets stuck. Via web I get the playback screen but no timeline and just a loading circle. On Android it gets stuck with the popup “communicating with server”.
How can I fix this issue?
Putting the Plex Library on the NAS doesn’t work well, if at all, as you’ve found out.
If I may recommend, a better choice is to place the metadata in /home/plexdata.
Now you have local filesystem locking which is infinitely better than ‘advisory locking’ on a network share. This is needed by the database itself.
You also have iNotify support (needed by the EAE codec). EAE doesn’t work without it.
Hi, thanks for your quick response.
since the recorded file is propperly shown in my plex library and the file itself is intact: where’s the exact difference between playing files recorded previous to the move, even if I have never played them before, and recently recorded media? Since the OS ran very low on free storage I’ve temporarily disabled plex generation of thumbnails and chapter previews. File locking should not be a concern, because the NAS folder is used by my plex user exclusively.
If I move my Library to /home, are there any recommendations or experiences using plex with thin provisioning? My media-os is running virtualized via VMWare ESXI 6.7 and I want to avoid to block storage beforehand unneccessarily or to monitor the vm’s storage needs manually outside of my automated host-level notification workflow.
NFSv4 (TCP based with automatic file locking) is the only one I know which works correctly.
SMB mounts on Linux are advisory locks. They aren’t hard (true) locks. (There is no guaranteed exclusivity with SMB mounts in Linux). This is what messes with the database and will corrupt it.
The files themselves will be fine. The transcoder, which is handling all the processing itself, doesn’t need to coordinate externally. Only PMS, with its multitude of asynchronous other threads, needs those locks.
In the end, I’ve decided to leave the plex library in place and to provide more storage using lvm and thin-provisioning-tools. In the long term, I might want to try and migrate my current setup to docker, which looks as if its even easier to maintain.
Thank you for your help and for sharing some insights!
