this post was submitted on 01 Jun 2023
46 points (100.0% liked)
/kbin meta
110 readers
1 users here now
Magazine dedicated to discussions about the kbin itself. Provide feedback, ask questions, suggest improvements, and engage in conversations related to the platform organization, policies, features, and community dynamics. ---- * Roadmap 2023 * m/kbinDevlog * m/kbinDesign
founded 1 year ago
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Has anyone tried creating a KBIN server using the Docker instructions since Saturday? I got a development server running on Saturday and I decided to bring up a production ready version on Sunday, but now I'm running into two errors while creating it.
First:
In these instructions:
$ sudo chown 82:82 public/media
$ sudo chown 82:82 var
$ cp .env.example .env
There is no directory "var" so the chown fails
Second:
When running docker compose build --pull --no-cache
It now fails with this:
=> [php app_php 20/21] RUN rm -Rf docker/ 0.2s
=> ERROR [php app_php 21/21] RUN set -eux; mkdir -p var/cache var/log; if [ -f composer.json ]; then composer dump-autoload --classmap-authoritative --no-dev; compose ...
This also seems to relate to the var directory, and so must be part of a recent change in the build that relates to the var directory.
Can anyone help?
Thanks
I have tried creating an instance on both Debian 11 and Ubuntu 22.04 with both Docker and the manual steps.
I can get the build to work fine. But at best I get 500 errors with no signs of logs as to why, not in postgres, not in nginx, not in redis or rabbitmq or even syslog.
On Docker i have both the kbin_messenger and kbin container boot looping with an error about creating cache directories. But since the containers are restarting I cant even get into bash on them to see why.
I finally said fuck it and ran the compose for "prod" and the containers start but again, 500 errors.
At this point....i give up i feel. I would love to host an instance and I am familiar with some of these modules (ie: NGINX, redis and postgres) but with how this is built, I dont see where the breakpoint is.
I operate my own homelab and have a background in SRE, so I figured I'd try out the same. I've wrangled a Mastodon instance install before, so this couldn't be too hard, right?
My approach started by using containers via lxc as a quick and dirty way of getting a development environment that enabled me to figure out if it worked and then see if I could then wrap it into a proper docker container and look at potentially publishing that.
On my first attempt, I did the manual route. I skipped the redis, rabbitmq, and postgres installs as I already operate those elsewhere on the network, but I got everything else running. Unfortunately, I also experienced the 500 errors. Most of the front page loaded, except for the content where the 500 error was displayed. Even with some digging around, I couldn't find a clear path to figuring out what was causing the 500 error, as the Mercure hub was seeing subscribers connect and disconnect. Gave up.
I then figured maybe the docker route might be easier/streamlined. I'm not a fan of duplicating services, but I thought that if the core workflow was solid enough, I could put effort into splitting them apart and go from there. Unfortunately, I don't even get past the docker-compose build for dev. Docker compose hangs forever.
Yeah. Kbin install made me eat some humble pie for sure. I think someone called me a normie in a lemmy thread describing my troubles. Lol.
To be fair some of the parts like mercure and rabbitmq im unfamiliar with. But it was a stone cold stumper for me and that’s rare. I’m fairly familiar with Linux admin, and even some of the tooling like docker. But I just couldn’t get it to work in a cohesive way. I ran plenty of Linux servers from Drupal instances, Postgres, nginx for all sorts of shit, etc etc.
My lemmy instance took….45 minutes to roll out. Though I already had an Ansible box sitting in my lab.