https://github.com/openshift/microshift/pull/1881#issuecomment-1579262622:

If the cp fails, then recovery is ... rm -rf /var/lib/microshift

Removing the files as an automated response to an error could potentially remove some of the information needed to figure out what the error happened in the first place - another move operation might be safer (assuming that running out of disk space isn't the error which caused the problem)

Current implementation:

• Rename /var/lib/microshift to /var/lib/microshift.saved
• Copy /var/lib/microshift-backups/backup-name to /var/lib/microshift
• If fails: remove /var/lib/microshift and rename /var/lib/microshift.saved back, exit with error
• Remove /var/lib/microshift.saved

Questions:

• Should we change how copy fail is handled?
• What's better from supportability/debuggability point of view? Maybe output of cp --verbose is enough?
• What's better from "make it as automated as possible" (aka require a little manual intervention as possible) perspective?