[Discussion] Suggestions about Docker Swarm

[coppolino97]

Hello folks,
During the latest day I deployed in my lab, for the first time, docker swarm.

I resume my tiny deployment:
- three VM in same LAN
- shared storage using GlusterFS. The storage is shared by VM theirself.
- one VM is docker swarm manager, other two virtual machines are docker swarm workers.

I decided to use GlusterFS to replicate docker data on each node of my environment.
Right now I deployed a very simple wordpress website (frontend+mysql backend) with success.

How did you manage container deploy? Do you usually deploy the same software on each node via "Replica" option?
I deployed docker on just first worker, so every time I turn off a worker node, containers take time to start on the secondary worker.
Moreover I need to change IP in my URL to still surf website.

Do you use different storage solution?
Do you have other suggestion for me?

Thanks a lot
Federico

[Delobian]

In my experience, I usually go with the "Replica" option for deploying containers, especially for frontends. It helps ensure redundancy and high availability. Also, I tend to store static files separately, often using an FTP server.

For deployment automation, I often leverage GitLab's CI/CD pipelines. This way, with each build, the deployment process kicks in automatically, ensuring seamless updates. And having replicas definitely helps in achieving zero downtime during production deployments.

------------------------------
Délobian - [moderator removed]