Ive got a lot of downloaded YouTube videos. I used to play them through Plex back when I ran that, but once I switched to Jellyfin there wasn’t a good way to set them up at all. So I was super excited when I found TubeArchivist. All the posts I saw on it seemed to be positive, so I set it up yesterday in Docker.
It seemed to work alright at first, but it took a really long time for the server to start up once I started running the containers. I’m also running Navisphere, Jellyfin, Kiwix, and a Tor snowflake node, which all booted up instantly, but this took 5-10 minutes to be accessible. It also took like twice the Docker resources of all my other containers just to run.
Doing anything in it was super slow too. Logging in was unresponsive, took a couple minutes to load. There was no feedback either so I thought it was just frozen. I had all my videos I had downloaded previously, so I tried importing them and that just brought my system to a complete crawl. It maxed out Dockers allocated resources, even exceeding the CPU allocation. It also caused Docker to crash twice.
Does anyone have any experience with this program? Did I do something wrong in my docker-compose file (I just modified the example in the github repository) or is the program just super poorly optimized? I’ve been able to run all the other containers on my system just fine. It has a lot of the same functionality as Jellyfin, which does not do this kind of crap. I was so excited but its been a huge letdown. Any thoughts?
You must log in or register to comment.
I tried it alongside Pinchflat and ended up settling on Pinchflat.
Why? I’m not going to say TA is bad, it worked well, but Pinchflat’s logic works better for me. Also, I was using TA with an existing Elastic search cluster. When I switched my logs to Loki/Garage I decided not to keep the ES cluster around just for TA.
I took a look at pinchflat but it seemed more focussed on the archival side of things rather than the playback side. Would you say thats accurate? Because honestly all I’m looking for is a frontend to easily watch my videos. I only download an odd video here and there and for that purpose yt-dlp works perfectly fine.
I only tried it for 5mins and then switched to TubeSync and never looked back
How’s the playback/library organization for TubeSync? That’s my primary interest with TubeArchivist.
I had the same issue when testing it, my server wasn’t powerful enough to run it. Elasticsearch seems to be the main culprit since its pretty heavy on both CPU and RAM, and that’s an intentional choice according to the developer: https://docs.tubearchivist.com/faq/#why-elasticsearch
Don’t have the same problems as yours. My only issue with TA is that it’s very picky about the ES and Redis version, so currently I pin them to a specific build. I do worry that one day TA will be updated so much that it won’t support older builds of ES and Redis and my whole thing may break.
But when it works, it works perfectly, giving Jellyfin all the metadata to display them neatly (with the TA plugin of course).
You guys are giving me hope, I just have no idea where to begin. Lol.
How did you set them up with Plex?
I tried TubeArchivist a while back but as I recall it was much more about archiving entire channels and didn’t have great customization like “Only keep latest two episodes”. It might have changed since then.
Its been a minute since I’ve used Plex so I don’t remember the exact specifications, but I set up a separate library in Plex and directed it to my YouTube folder where I had everything pre-organized, and then I selected the option in Plex that was something like “show folders,” which organized the Plex library the way I had the directory organized. It wasn’t perfect because I had to set up everything manually, but all I wanted was to be able to watch my YouTube videos from anywhere in the house, and for that it was perfect.
I run TA with success since many months. It’s fast and responsive, so I guess it’s an issue with your setup?
The metadata bridge to jellyfin alfo works pretty nice, I have no complaints…
You can check my notes at https://wiki.gardiol.org/doku.php?id=services%3Atubearchivist
But be advised I am on rootless podman with docker compose
Okay, your notes have definitely helped a bit. I compared your compose file against mine and removed the HOST_UID, HOST_GID, and the healthcheck sections from the default file. Its still taking an abnormal amount of time to boot up and login, but the container isn’t completely throttling my system anymore. I suspect the healthcheck was the culprit. A step in the right direction though, thank you!
I removed health checks because o think they don’t work properly with podman how I use it, but I might be wrong.
Anyway, glad it helped! That’s the spirit why I wrote it.
