Open ergonlogic opened 9 years ago
Note that this should be independent of whether a site is using a bucket for it's files.
Also, ideally we'd stream backups directly to a bucket without touching the local filesystem. I'll have to look into how we do backups currently though. We might need to generate a tarball locally, move it to S3, then delete the local copy.
It should be sufficient to provision a backups bucket per client, but we might also want to allow it per site.
Since we don't run any back-end ops when we save a client, we'll need to provision the bucket in the front-end. This will probably entail moving a couple more functions into our common include. We'll then need to write this bucket name (and credentials?) into the site context.