I would like to use the thanos swift cluster as storage back-end for chartmuseum (https://chartmuseum.com/docs/#using-with-openstack-object-storage). If the S3 API is a better choice, that's fine too.
- ChartMuseum will be deployed active/active in both DCs.
- ChartMuseum will act as proxy to swift, so no direct client connections.
- Currently we have ~291 objects with a total of ~3.5MB which I expect to slowly increase.
- I/O will be mostly reads, writes will happen only on push of a new chart.
- The data that will be stored can be (with some effort) regenerated from git, but swift will be the "authoritative" storage for the artifacts (chart tgz's).