Sequential restart when using docker-compose up with scale

[corradio]

Hello,

It seems to me that when scale is > 1, and a new image is being pulled, using docker-compose up restarts all containers in parallel rather than sequentially.
Doing a cascading restart would ensure zero downtime.

Am I correct in my analysis?

Thank you,

Olivier

[shin-]

Both are valid strategies depending on your use-case, but you're right that we don't currently offer an automated way to do zero-downtime deployments. In the meantime, you may want to take a look at #1786 (comment) which outlines how to do so manually.

[corradio]

Hi @shin- ,
Unfortunately your method won't work as I have more than 1 container in production.
With previous versions of docker/docker-compose, it seems like containers were restarted sequentially. Did anything change?
It would be nice to be able to control the start strategy in docker-compose (sequential vs parallel)

[shin-]

With previous versions of docker/docker-compose, it seems like containers were restarted sequentially. Did anything change?

Unless you're looking at versions from 2 years ago, that wasn't the case.

[corradio]

I was indeed working with a very old version (due to CI constraints).
Any idea if it's possible to add the feature I'm requesting? I'm up for contributing if needed.

[domq]

Replying to the original question, it is still not possible to do sequential restarts. However, you can prevent docker-compose up --scale from restarting existing containers using the --no-recreate flag. This is enough to do zero-downtime restarts if you use only run one serving instance (by momentarily scaling to two, then deleting the oldest container).

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically closed because it had not recent activity during the stale period.