Feature Request: Support Rolling Updates (one-by-one container restarts)

[CngTZ55]

Hi! First of all, thank you for this project — I’m using WUD as part of my CI/CD workflow and it works great!

My setup is the following:

• When I push a change to the dev branch
• GitHub Actions builds and pushes an image to my Harbor registry
• WUD watches the registry for updates
• When a new image tag appears, WUD redeploys the containers

This workflow works perfectly.

However, I have a use-case where WUD currently falls short: rolling updates.

Right now, when an image update is detected, WUD restarts all replicas of a service at once.
If a service has multiple replicas (e.g. replicas: 3), this results in full downtime.

I would like to request a feature similar to Watchtower's --rolling-restart, where WUD could:

• Update one replica at a time
• Wait for the updated replica to become healthy/ready
• Then continue with the next replica
• Until all replicas are updated

This would allow zero-downtime deployments and make WUD viable for production workloads where availability matters.

A simple flag such as:

WUD_TRIGGER_DOCKERCOMPOSE_<trigger_name>_ROLLING=true

(or any name you prefer)

would be extremely useful.

Thank you for considering this feature! Let me know if more details or examples are needed.

[maskofsanity666]

Excellent idea, seconded, would be very helpful!