So, in the process of recovering from a disk-full situation during an otherwise normal Discourse upgrade, I realized that we are apparently using the “devicemapper” Docker storage driver, which seems to be heavily discouraged for Discourse, as I gather from other threads here.
What would be the recommended procedure to fix that and use the recommended driver instead (aufs, right?) – backup/reinstall/restore, or is there an easier way? This is on an EC2 instance with AMI Linux 2016.03 and Docker v1.11.1.
And on a more general note: I stumbled upon this devicemapper thing pretty much accidentally; according to other threads, it shouldn’t even have been possible to set up Discourse this way? I may have missed a warning fly past, but I’m sure I did not deliberately override anything during the installation process. I set it up according to the guide on GitHub around the end of May. Let me know if there is anything I can do to help prevent this for others in the future.