Closed kcajmagic closed 5 years ago
Think of it as a primary cluster and a backup list. So if the primary cluster comes back, the active connection should go from the backup list back to the primary.
We need a configurable interval for when to ping the primary cluster to check if it's back. Also, I think we need an ordered list, otherwise different svalinns may fall back to different clusters, causing issues (only a problem if we are supposed to insert only to one region).
Yes, I agree.
Closing this as it is not needed anymore.
If the database cluster the package is talking to goes down, there should be a back up list of clusters to try.
Think of it as a primary cluster and a backup list. So if the primary cluster comes back, the active connection should go from the backup list back to the primary.