One of the things I like about Docker is being able to quickly/easily reset everything back to a known good state. Currently, if I enable autorestart it appears that the enrtypoint is re-run but the container is re-used. This means any local changes inside the container that happened on the previous run are still there (e.g., caches, temp DBs, etc.) and this can sometimes mean that the container gets stuck in a failing state.
How can I make it so my container will autorestart if it exits with failure, but does so in a new container? Autodestroy is apparently incompatible with autorestart, which is what my gut tells me the right option is.