Amazon S3 sync theory

by dKen   Last Updated January 11, 2019 12:00 PM - source

I'm setting up a backup of my vital data (database and images, mainly) to S3 from my Ubuntu box. I've set up Amazon S3, and installed S3FS to mount the bucket on my machine. This is working well, but when I test my script, I'm getting a warning:

s3fs: MOUNTPOINT directory /aaa/bbb/ccc/ is not empty.
s3fs: if you are sure this is safe, can use the 'nonempty' mount option.

So, I'm not sure of the theory of how the mounted drive should work. I was thinking that I could mount the drive just before I put a new file into the directory to back up, and then unmount the drive when I'm done. Is this flawed logic? Should I just mount the drives once and leave them? If the mount fails for some reason and I need to re-mount, should I be emptying the directory before re-mounting?

For additional information, I'll be overriding database backups on a rotational basis, and rsyncing a large directory of images.

Any help with the theory or ideas for how I can set this up effectively and by using best practice much appreciated!

Edit: Or would s3cmd be better suited to my backup plans?



Related Questions


s3fs: how to force remount on errors?

Updated January 10, 2019 00:00 AM

AWS S3 no space left on device

Updated January 11, 2019 11:00 AM

How to make s3fs work with IAM roles?

Updated October 07, 2015 08:00 AM


S3fs performance improvements or alternative?

Updated October 17, 2015 05:00 AM