Open davepacheco opened 1 year ago
Alternatively: if we know a given Nexus is definitely not running (and especially if it will never run again), we could have an operation that assigns its sagas to another Nexus instance running the same image. This seems important for the sled removal case, where we can't actually drain old sagas. You could think of this as a simple, constrained, controlled case of failover.
RFD 289 describes a few pieces required for Steno upgrade: