Monitoring fuse mounts

[JohnStrunk]

Describe the feature you'd like to have.
The CSI pod should be able to monitor the fuse mount processes and report on abnormal conditions.

What is the value to the end user? (why is it a priority?)
The CSI driver will spawn a FUSE process as a part of each volume mount. Currently, these extra processes are just "fire and forget." However, it is possible for one of these fuse processes to crash. Even with #3 this may not be detectable. A daemon within the pod should watch the fuse processes and ensure that if one crashes, a suitable error is logged so that an admin can easily diagnose why a pod has lost access to storage.

How will we know we have a good solution? (acceptance criteria)

• An error message is logged to stdout (and visible via kubectl logs <container> when a fuse process abnormally exits
• No error is logged if the fuse process exits due to an unmount

Additional context

• Bonus points for detecting and logging loss of connection to the server (that may reconnect in the future). <== I suspect this will be visible with mount flag to log to stdout to be provided #3
• It would be nice if there was a way to automatically remedy the crash, but I'm currently at a loss.