I am using a linux-server for Plex. My media all is stored on a s3 object storage. To set it all up I followed https://www.scaleway.com/en/docs/configure-plex-media-server-with-object-strorage/ and it worked properly.
But - yes now comes the “but” - I have a Problem:
After several hours the media isn’t shown as available anymore on Plex. If i try to browse the mounted s3 object storage (edit my libraries), it seems to be empty (or unmounted). But then, when i connect to my Linux-Server and list all content (with ls) of the mounted storage, then it is there. And furthermore plex is then able too see it again too and I can play the media again.
Same thing happens over and over again.
I use ubuntu 18.04 LTS and the all software used is up to date (plex, s3fs, …)
Oh damn, I just figured, it seems to corrolate with being logged into the Server or not. So I think I will try to find out, how being connected affects the mounting, and first and foremost, how to Prevent unmounting.
Double check how you mounted those drives.
I have no 1st hand experience with s3 storage and how to mount it in Linux… but from what you describe it sounds like the drive is mounted but not for the plex user account used by Plex to access your media.
I’ve never attempted mounting an S3 bucket like this in a Linux server, but I am an AWS Certified Solutions Architect. My advice to you is to look at using a Storage Gateway, also a service provided by AWS. This allows an S3 bucket to be mounted as a filesystem and interacted with as such. It’s intended purpose is for hybridizing a local datacenter’s file server with AWS resources. This might provide a more stable stable connection to your bucket. Just a thought.
Another thing you can try is using AWS Edge Locations (essentially cache boxes) to provide access to your files via pre-authenticated URLs - this would restrict the access instead of leaving it globally public. Also, something you’ll have to research to modify to your purpose. Note: data transfer between S3 and Edge locations is free, so if your library were to be shared and a popular show watched by several users you would not incur a data transfer charge for each GET of the file, only for the original GET.
PS - on S3 you can set a lifecycle policy that will move files from the most available and expensive tier of S3 into S3-IA and more. This can be based on the last date accessed it and will reduce the cost of storing files that have been sitting idle for quite a bit of time. This lifecycle policy can move a file all the way into Glacier but data transfers out of Glacier can be pricey compared to S3-IA depending on the usage. Plex scanning may also make this bit of S3 magic more expensive so you’ll need to evaluate if it suits your purposes.
Hi there
Thanks for all your support. Actually i looked into the behaviour, that it was mounted, as long as I was connected with root.
So actually i moved the s3fs.service file to etc/systemd/system and changed {HOME} in passwd_file to a correct link, since {HOME} was rejected by then.
This did the trick. Since then I have a absolutely stable plex experience without any problems.
After all I’m happy with my solution, since I had serious storage size problems before, which are gone by now. My obect storage is quite cheap and no cost for traffic. So happy streaming for me. It is just slightly slower in starting a movie, Maybe 1s or so.
