4 ms·

Interesting, from the doc[1]: "Because of various issues Invidious must be restarted often, at least once a day, ideally every hour." [1] https://docs.invidiou

by mpitt 3y ago

Interesting, from the doc[1]: "Because of various issues Invidious must be restarted often, at least once a day, ideally every hour." [1] https://docs.invidious.io/installation/#post-install-configuration https://docs.invidious.io/installation/#post-install-configu... I guess it's related to this issue, a memory leak (or probably multiple): https://github.com/iv-org/invidious/issues/1438 https://github.com/iv-org/invidious/issues/1438 Also interesting that the issue is almost 3 years old and mentions even older reports.

RockRobotRock 3y ago
Getting invidious to run was a gigantic pain for me. It definitely consumed all my RAM and crashed my system at least once.
- turtlebits 3y ago
  Odd, I have the docker image running on a cheap 4gb VPS and it's been rock solid so far (set it up about a week ago).
- toastercat 3y ago
  It was definitely a pain in the ass before the official docker image. Have you tried that?
  - RockRobotRock 3y ago
    Thank you! I'll try the Docker image.
    - toastercat 3y ago
      No problem. And I meant to mention the Docker Compose config they have here [0]. This is what I use to self-host on my Raspberry Pi, and it works great. Just make sure to add a cron job that restarts the container every hour to get around the memory leak issue (my cron job is literally /bin/docker restart <container_name>) [0] https://docs.invidious.io/installation/#docker https://docs.invidious.io/installation/#docker
- Karupan 3y ago
  As long as you aren’t hosting a public instance, the docker config in their docs should work fine. As for RAM consumption, I have set limits, so not sure if it’s actually happening.
- unixfox 3y ago
  You mean when compiling? It's normal. There are recommendations about the memory needed for compiling. If you do not have enough RAM just use the docker image which is already precompiled.
Helmut10001 3y ago
I don't observe this. Invidious was super easy to setup (docker-compose.yml) and I never restart. Has worked flawlessly since 2 years.
- mpitt 3y ago
  Huh, could be that the doc is outdated then. Thanks for the info, maybe I'll give it a try after all!
  - Helmut10001 3y ago
    I verified and I did not see a drop in memory consumption after `docker compose down`. Added this to the issue you mentioned.

by mpitt 3y ago

RockRobotRock 3y ago

Getting invidious to run was a gigantic pain for me. It definitely consumed all my RAM and crashed my system at least once.

turtlebits 3y ago
Odd, I have the docker image running on a cheap 4gb VPS and it's been rock solid so far (set it up about a week ago).
toastercat 3y ago
It was definitely a pain in the ass before the official docker image. Have you tried that?
- RockRobotRock 3y ago
  Thank you! I'll try the Docker image.
  - toastercat 3y ago
    No problem. And I meant to mention the Docker Compose config they have here [0]. This is what I use to self-host on my Raspberry Pi, and it works great. Just make sure to add a cron job that restarts the container every hour to get around the memory leak issue (my cron job is literally /bin/docker restart <container_name>) [0] https://docs.invidious.io/installation/#docker https://docs.invidious.io/installation/#docker
Karupan 3y ago
As long as you aren’t hosting a public instance, the docker config in their docs should work fine. As for RAM consumption, I have set limits, so not sure if it’s actually happening.
unixfox 3y ago
You mean when compiling? It's normal. There are recommendations about the memory needed for compiling. If you do not have enough RAM just use the docker image which is already precompiled.

Helmut10001 3y ago

I don't observe this. Invidious was super easy to setup (docker-compose.yml) and I never restart. Has worked flawlessly since 2 years.

mpitt 3y ago
Huh, could be that the doc is outdated then. Thanks for the info, maybe I'll give it a try after all!
- Helmut10001 3y ago
  I verified and I did not see a drop in memory consumption after `docker compose down`. Added this to the issue you mentioned.