Does --rolling-restart restart one by one replicas ?

[MaxR522]

I am running an application with 2 replicas, and I want to ensure zero downtime when upgrading the Docker image. Specifically, I want Watchtower to update the replicas one by one, so that at least one replica remains running at all times.

The documentation states:
"Restart one image at a time instead of stopping and starting all at once."

However, since my replicas are based on the same image, it's unclear to me whether the --rolling-restart option restarts each replica individually or if it still results in downtime for the entire service. Could you clarify if this option ensures one-by-one rolling updates for replicas?

services:
  webapp:
    image: myapp:latest
    deploy:
      replicas: 2 

[kinduff]

Can confirm it does. It will go one by one, can be seen here

watchtower/internal/actions/update.go

Lines 97 to 121 in 76f9cea

	func performRollingRestart(containers []types.Container, client container.Client, params types.UpdateParams) map[types.ContainerID]error {
	cleanupImageIDs := make(map[types.ImageID]bool, len(containers))
	failed := make(map[types.ContainerID]error, len(containers))
	for i := len(containers) - 1; i >= 0; i-- {
	if containers[i].ToRestart() {
	err := stopStaleContainer(containers[i], client, params)
	if err != nil {
	failed[containers[i].ID()] = err
	} else {
	if err := restartStaleContainer(containers[i], client, params); err != nil {
	failed[containers[i].ID()] = err
	} else if containers[i].IsStale() {
	// Only add (previously) stale containers' images to cleanup
	cleanupImageIDs[containers[i].ImageID()] = true
	}
	}
	}
	}
	if params.Cleanup {
	cleanupImages(client, cleanupImageIDs)
	}
	return failed
	}