QA

Topics to consider:

• Host machines ephemerality
• Need of backups and recovery
• User experience

The host machine is set up manually and in case of a migration or recovery a infrastructure as code approach would enable less downtime. Perhaps a terraform+ansible setup could be considered.

The container images for the frontend and backend of retrospected are not available anymore. The upstream images require a license. If we lose the local images and don't have an external backup, we can't replicate our current setup. Best case would be to own a license, second best is to host the images on a suse controlled registry, easiest solution would be to save the images as tar files somewhere team accessible.

Currently no reverse proxy is used for services on the host machine. Access to services is done via port annotation in the urls. If only one service should be accessible via the standard http port 80, then that can easily be done via docker port mapping. Best solution would be a local reverse proxy, like Traefik, with a configuration that enables auto forwarding of ports using docker labels.

The hostmachine/dns record mention core, it suggests more services hosted by the qe-core team. If that is wished in the future, we may consider to increase the size of the host machine. For now it is only at 1cpu and around 4gb ram.