• 0 Posts
  • 12 Comments
Joined 3 years ago
cake
Cake day: June 9th, 2023

help-circle



  • I work per stack. Each stack can have multiple services/containers. So one db per stack. I usually work with the compose file provided by the app I want to run and they usually provide their database container in those. My persistent data for a stack is in a single folder for every services. For backup, I only have to backup the compose and environment files and also the persistent folder and that’s it. Every stack is independent from each other so it’s easier to move them from a server to another if I want to use more than one server.

    I used to use only one db container for all db. It worked well, until I had a data corruption issue. I had to rollback from a backup for all services that used this db. Then I moved to 1 db per stack and it happened again, but this time, only that stack was impacted, so no big deal. Just did a rollback and everything was fine.



  • To manage my docker containers I migrated from and to multiple tools. I started directly with Unraid with individual containers. Then I discovered docker compose and stacks. So I migrated to portainer. Splitting the postgres and mariadb databases were the worst thing in this. Then I tried dockge, so migrated from portainer to that. It lacked some portainer functionality, but was fine for a short while.

    Finally I discovered Komodo and migrated to that as it’s FOSS. Now I’m only on Komodo, I have no server limit (5 on portainer), had all the functionality of portainer and more. It is my go-to docker manager and the one I recommend to everyone. The only “issue” is that you need to write the name of the stack and click confirm twice if you want to redeploy, restart, stop, destroy. Or you have to click confirm twice after saving when you edit your files. Not much of an issue and it can also protect from accidental clicks. My next step will be to migrate my stacks compose and env files from UI defined to files on the server that I can sync.