flor0

joined 1 year ago
[–] flor0@sh.itjust.works 3 points 7 months ago

Seems similar to what I'm doing but it uses the nextcloud:fpm-alpine image, which is community maintained. It's a valid approach but I prefer to tweak my php-fpm config myself. By using their docker image you basically lose a bunch of freedom to configure your setup, but it's easier to set up for sure

[–] flor0@sh.itjust.works 2 points 7 months ago (1 children)

Thanks for the kind words! Could you link me to the apache container you use? I would like to replace nginx since it's not officially supported but still need to do some research on apache

[–] flor0@sh.itjust.works 1 points 7 months ago (1 children)

Alright, thanks a lot! You probably saved me a lot of headache in the future

[–] flor0@sh.itjust.works 1 points 7 months ago (3 children)

True. That's a good improvement. Thanks a lot! By the way do you know if the nextcloud trusted_proxies setting allows for hostnames such as "traefik" as compared to just IP addresses?

 

For years I've been running my Nextcloud on bare metal. Now I finally decided to switch it to docker. I looked into Nextcloud All-In-One and it seems easy to deploy but it lacks configuration options like multiple domain names and putting it behind a traefik reverse proxy is kind of a pain to set up.

You can check out my solution on GitHub. I decided on a docker compose setup with nginx, php-fpm and redis(redis is now replaced by KeyDB) in separate containers. Obviously it's for experts but it's a lot more configurable. than AIO. It's also just as easy to migrate to as with any bare metal setup and just as configurable.Yes it's still a pain to set up, but better than the bare metal version lol

What do you guys think about putting the different components (webserver, php, redis, etc.) in separate containers like this, as compared to all in one? Feedback is greatly appreciated!

[–] flor0@sh.itjust.works 4 points 1 year ago

I think yandex does its own thing, too. I agree it's very hard to compete with google when they have had so much time to improve their algorithms and infrastructure