I tried searching the forums, but didn’t find a (solved) issue that matches mine exactly.
Generally, there seems to be a lot of confusion about getting collabora to work, and from what I understand, it’s mostly due to some ssl-certificate complications. My setup is:
Nexcloud on a hosted storage - no access to the underlying machine (no ssh login or console), but fairly vanilla nextcloud (24.0.5). Hosting company (hetzner) manages update of NC as well.
My own domain name for NC, hosting company takes care of SSL.
Collabora running in docker (configured with docker-compose) on a separate rented virtual machine (same hosting company), full access to everything on that machine.
In front of collabora, an nginx reverse proxy with letsencrypt ssl cert.
ssl confirmed working - tried switching out collabora docker for a trivial web-api, I can call it over https using my domain name without problems.
Calling https:///hosting/discovery yields an xml file revealing the guts of the collabora install
I have tried a lot of different settings for nginx.conf and collabora docker without success, getting different errors in the logs (mostly ssl related). I think what is confusing me is that collabora apparently takes care of ssl on its own somehow, without needing the ssl-functionality of nginx - I don’t fully understand how that works.
Since the problem is sitting in front of the keyboard, I’m not going to post my nginx configs - that might just be embarassing.
Instead, I would like to ask if there is a standard configuration for this setting - collabora in a docker on domain A, NextCloud “somewhere else” on domain B communicating over https with each other and the user? I have found the Collabora documentation to be quite tight-lipped on the details, I ask here since I suspect there are more people here that have actually gotten their hands dirty with some real working setups.
I have seen some configurations of collabora using traefik instead of nginx, with some quite elaborate ssl-configuration commands. Is this somehow better supported - should I maybe switch to traefik?
Thank you for your reply.
As usual, while writing my problems down in the original post, things became a bit clearer (“talking to the rubber duck…”).
Thing is, I had tried out several different nginx.conf settings. Searching the web for this problem results in 2 or 3 slightly different such configs, which is a bit suspicious in itself, each setting resulting in “almost-working-but-yet-not-quite” with different things broken for each setting (only constantly broken thing over all of them was my self-esteem…).
So I ditched nginx, and tried out treafik instead, which seems to make do with even less verbose config. This thread has excellent write-up of this. The thread describes NC and collabora running on dockers side-by-side, but using just the traefik and collabora bits works perfectly for the case when the nextcloud server runs somewhere else.
So apparently, my original problem was a wrong configuration of routes in the nginx config file. But since the traefik-based setup works, I’m not going to try to find out what I messed up…
For reference, this is the docker-compose setup that ended up working for me :