This application is designed to be used in meat space, and not used by QE.
Requires a full PRD but...
- QE check certain files exist, which means accessing a TAR file which is nasty
- If we build a JSON back end, then we can basically have a machine readable data structure which is ordered by namespace, resource type, resource, file. And contains all the same information.
- They cannot access log volumes as this is interactive.
- Again with a JSON backend we can communicate which logs can be collected, then allow them to be collected via the CLI on a second pass.
- This will apply to normal pods and log volumes, this streamlines the process, also provides auto download for all classes of log.
- They download all logs regardless
- It is naively just screen scraped based on pattern matching (ugh)
- We should make (non) redaction optional and only download the correct log.