Using volume mounts vs bind mounts for the IoT Edge system modules

[anttikes]

The documentation instructs using bind mounts for persistently storing the data of edgeAgent or edgeHub modules on the host filesystem.

Docker's own documentation recommends using volume mounts instead.

Is there any clear reason why bind mounts are preferred in the documentation? Would the same benefits (i.e. robustness) be achieved by using volume mounts?

In what kind of a situation would it be necessary to manually manipulate the files persisted by the system modules from the host side?

[mikanygTC]

We have chosen to ignore the documentation and opted in for using volume mounts as per the recommended docker approach.

I have been following Azure IoT Edge since 1.0, and I think the docs where written a long time ago, and might haven't been updated to reflect dockers recommend approach.

bind mounts mostly makes sense if you need to be in control of the folder on the host, either by sharing with other containers or daemon on the host. Otherwise I would go with volumen mounts.