Does running docker commit on a running container produce a consistent, point-in-time image?
Are there disadvantages to doing this regularly? Does it introduce any incremental overhead in terms of the storage driver?
(for the record, I am aware of volumes and data-containers and the best practices regarding backups with docker, I’m just trying to understand how commit works)
I don’t know if it is atomic – that would take a careful code review. So to be on the safe side, I would recommend using docker pause on a running container before docker commit, then docker unpause.
If you end up doing the careful code review, please let us know what you learn!