Establish process/procedure for rebuilding old images

[lmsurpre]

Currently, we build and push our Docker image to DockerHub as part of our release process.

This works, but ends up leaving vulnerable images because the underlying layers get patched but we never re-build our images on top of them.

What we should be doing is rebuilding the images at some interval so that our image is always up-to-date at the lower levels (OS, Java, Liberty, etc), probably via either autobuilds or GitHub Actions scheduled events

As a forcing function, Docker recently changed their terms of service and will start deleting stale images that haven't been rebuilt in the last 6 months.

[prb112]

Team to setup a side discussion on how to approach

[prb112]

Proposal

• Rebuild on a GitHub Actions - schedule
• All Tags which have a corresponding docker image (>= 4.1.0)
• 4.2.0-FIXED is a special case

Go from there.

[lmsurpre]

Rebuild on a GitHub Actions - schedule

For the server itself, our base image is re-built daily and so that is probably a good schedule for us to follow as well.

For the schematool image (or any others), I'm not sure how often the base image changes.

[lmsurpre]

now that it only rebuilds the latest (4.11.0), we finally have an automated image rebuild!

first success: https://github.com/IBM/FHIR/actions/runs/2200236782

[d0roppe]

This seems to be building correctly now. Closing issue.