43
Is dockStarter a waste of time? (lemmy.ml)
submitted 2 years ago* (last edited 2 years ago) by to c/selfhosted@lemmy.world
30 comments fedilink hide all child comments
 

Should I be learning docker compose instead of relying on dockStarter to manage my containers? I got portainer up, should I just use that to manage my stack?

I’m committed this summer to finally learning docker. I’m on day 3 and the last puzzle piece is being able to access qbittorrent locally while running the container through the vpn.

you are viewing a single comment's thread
view the rest of the comments
[–] 8 points 2 years ago (10 children)

If you're learning in any kind of professional capacity, you may want to get familiar with running things on k8s. I would never deploy Compose in any kind of production environment.

  • permalink
  • fedilink
  • source
  • hideshow 10 child comments
    •
  • [–] 9 points 2 years ago (9 children)

    I would never deploy Compose in any kind of production environment.

    May I ask you why?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 9 child comments
    •
  • [–] 5 points 2 years ago (8 children)

    Aside from the myriad issues it has on its own, the easiest answer is: it doesn't scale on multiple machines and instances.

    Example: I have 10 services in a compose file, and I need each service to scale independently across multiple servers. Which is easier, more reproducible, and reliable: controlling the docker compose state across many instances, or communicating with a central management service with one command to do it all for me?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 8 child comments
    •
  • [–] 6 points 2 years ago (5 children)

    What you need then is swarm compose, that can run any service in global mode (in all nodes all the time) or scale mode.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 5 child comments
    •
  • [–] 5 points 2 years ago (4 children)

    No, then you're just orchestrating the service level stuff, and nothing else. Docker's tools will never compare to cluster scaling efforts where the entire horizontal layer to be scaled can be orchestrated from the instance up to containers.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 1 point 2 years ago (3 children)

    not sure I understand you, in docker swarm your containers are started on n number of works from a single compose file on a manager. you can add any number of work nodes to scale your service as needed

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 2 points 2 years ago* (last edited 2 years ago) (2 children)

    N number of EXISTING nodes. Proper container orchestration platforms handle all the provisioning of instances, scaling of services, IAM...etc.

    Docker Swarm (and all Docker tools) only handle...Docker.

    Swarm is another thing I would never recommend in production.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 0 points 2 years ago (1 child)

    That's not container orchestration, that's infrastructure orchestration. Depending on your use case docker swarm could just the right tool for the job.

    You've been using Aws and they will happily let you add more nodes to your container runner of choice

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 3 points 2 years ago (1 child)

    Sure, but what you are describing is the problem that k8s solves.
    I've run plenty of production things from docker compose. Auto scaling hasn't been a requirement, and HA was built into the application (so 2 separate VMs running the compose stack). Docker was perfect for it, and k8s would've been a sledgehammer.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 2 years ago

    K8s isn't the only container orchestration platform out there, it's just what is the widely used flavor right now. Any of the micro clusters would still be better than the Docker tools, for a multitude of reasons, and if someone is learning about this right now, they might as well put the effort forth to get familiar.

    I've never seen a large scale Compose or Swarm cluster, and wouldn't be working for a team who ran such things. Alternatives would also be: ECS (if on AWS), Openshift, Rancher, and most other cloud platforms have some form of their own that handles provisioning, as well is IAM/RBAC seamless integrations, and other networking integrations for whatever platform.

  • permalink
  • fedilink
  • source
  • parent