Soliciting feedback and pain points in multi-node Swarm setups

[m90]

A lot of issues (e.g. #425, #329, #478, #423) that are openend in the recent months revolve around shortcomings of this image when using it in Docker Swarm setups that are composed of multiple nodes. Recently, I have spent some hours looking into the topic(s) recently and would like to gain a better understanding of:

• how many users actually run a multi node swarm cluster
• what are the shortcomings they face when using docker-volume-backup
• what are the workarounds in place to fix these
• what are the tools that are used instead

As far as I understand the current problems when using docker-volume-backup in a multi-node swarm setup are:

• No way to deploy a single service and backup data from all nodes, instead each node needs to host a dedicated service with dedicated configuration
• When deployed on a worker node, the image is not able to invoke custom commands in other containers
• No way to execute commands in containers that don't share the host node with docker-volume-backup

I do believe Docker Swarm is here to stay, so I think improving the situation is time well spent. However, time is also limited, so I would like to make sure I spend it on things that actually make a difference. In case you'd want to help out, please answer the following: As a Docker swarm user that deploys a multi-node cluster, what would be the single change in docker-volume-backup that would make your life a lot easier. If your idea is already very specific, feel free to propose a way of configuring the feature you describe.

Thanks, I am looking forward to read what you're coming up with.