I’ve been using Portainer to manage my homelab stacks from a single dashboard, which is more convenient than the CLI, but I’m not very satisfied with it so I’ve been looking for alternatives.
Portainer often fails to deploy them and is either silent about it, or doesn’t give me much information to work with. The main convenience is that (when it works) it automatically pulls the updated docker compose files from my repo and deploys it without any action on my part.
Docker Swarm and Kubernetes seem to be the next ones in line. I have some experience with K8s so I know it can be complex, but I hope it’s a complexity most paid upfront when setting everything up rather than being complicated to maintain.
Do you have any experience with either one of these, or perhaps another way to orchestrate these services?
I use Portainer, never had any issues with it and I’m running around 40 Stacks at any given time. There’s also Yacht, which is nice, but not quite as feature rich as Portainer but it’s super easy to use in comparison.
Currently it’s in the CLI, I just split my compose files in different concerns, and just use a bash alias that uses a wildcard to call them all.
But now as I’m adding a RPi in the stack to add some monitoring and a few light stuff, I’m also thinking of going to Kube. But as you say, it may be tough ^^
I store my Docker Compose files in an internal-only git repo (hosted on Gitea).
Drone is my CI/CD system, and I use Renovatebot to look for updates to container tags (never pull latest). My workflow is this:
Renovatebot runs once a night (at midnight) via Drone in a Docker container (I’ve written about this here).
If a new image tag is found, it opens a PR in Gitea.
I manually log in to Gitea and approve the PR.
The PR approval (merging to master) kicks off a Drone workflow that does the following:
Runs an Alpine Linux container
SSHes from the Alpine Linux container into my Docker host
Runs a script (on the Docker host) that basically runs git pull, then docker compose -f "$D" pull and then docker compose -f "$D" up -d.
This means I never manually update Docker Compose files, I let Renovate manage everything, I approve PRs, then I walk away and let the scripts run.
I also run a single-node K3s cluster that is hosted on GitHub. Again, using Renovate to open PRs, and I run Flux so watch for changes to master, which then redeploys applications.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !selfhost@lemmy.ml
A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don’t control.
Rules
No harassment
crossposts from c/Open Source & c/docker & related may be allowed, depending on context
Video Promoting is allowed if is within the topic.
I use Portainer, never had any issues with it and I’m running around 40 Stacks at any given time. There’s also Yacht, which is nice, but not quite as feature rich as Portainer but it’s super easy to use in comparison.
Currently it’s in the CLI, I just split my compose files in different concerns, and just use a bash alias that uses a wildcard to call them all.
But now as I’m adding a RPi in the stack to add some monitoring and a few light stuff, I’m also thinking of going to Kube. But as you say, it may be tough ^^
This is definitely an over-engineered setup…
I store my Docker Compose files in an internal-only git repo (hosted on Gitea).
Drone is my CI/CD system, and I use Renovatebot to look for updates to container tags (never pull
latest). My workflow is this:master) kicks off a Drone workflow that does the following:git pull, thendocker compose -f "$D" pulland thendocker compose -f "$D" up -d.I’ve written about step 3 here.
This means I never manually update Docker Compose files, I let Renovate manage everything, I approve PRs, then I walk away and let the scripts run.
I also run a single-node K3s cluster that is hosted on GitHub. Again, using Renovate to open PRs, and I run Flux so watch for changes to
master, which then redeploys applications.