I haven’t used this tool specifically, but anything involving Youtube is always going to be an uphill battle. In the instance of a video downloader like Channeltube, you frequently need to feed it fresh cookies to make sure Youtube doesn’t flag it as a bot. I assume this is a similar situation.

The requirements also list ‘‘2gb if you restart frequently’’ so there might be a memory leak issue, or the way it’s setup means the RAM usage will increase in perpetuity.

I would just script a task to restart the docker container every hour. Shouldn’t be too much a hassle, you’re unlikely to ever notice since it shouldn’t take more than a few seconds. I would track its RAM usage over a few days too. See if there’s something fishy happening. Something like Beszel will let you track your docker socket, so usage per container. Preferably left on read only imo.

I’d be surprised if most people running invidious for private use are following that advice. I’ve been on the master branch for a while and it gets updated fairly often. For my instance, I haven’t seen any reason to restart the service more often.

I stopped using Invidious a couple years ago, when I found out about FreeTube. I do have to update FreeT pretty often. I'll put Invidious back on my list. Thanks for bringing it back to my attention.

I haven’t had to adjust or look at my invidious instance configs in a long time, but I do remember that in their docker instructions and I think I configured my container to restart hourly (or maybe I made it every other hour? Daily?). I’ve had no problems with my invidious instance in a long time, but that may have nothing to do with frequent restarts.

That line predates invidious-companion. When the player and signature handling still lived in the main process, a periodic kick was the honest workaround, and the docs never really caught up after companion split it out.

What I’d push back on is the blind hourly restart, because it hides exactly the failure you just had. A companion image sitting a version behind shows up as things half-working, and a scheduled restart makes that go away for an hour at a time, so you never trace it. If you’re going to script something, script a check instead: hit an endpoint that actually plays a video, restart only when that fails, and log it when it does. Then a restart tells you something happened.

Small thing that would have caught your case: docker compose pull with no service name pulls all three, and docker compose images prints the digests you’re actually running.

healthcheck:
  test: ["CMD-SHELL", "wget -qO- http://localhost:3000/api/v1/trending >/dev/null || exit 1"]
  interval: 5m
  timeout: 10s
  retries: 3
restart: unless-stopped