- 최신
- 최다 투표
- 가장 많은 댓글
Unfortunately, no. The upgrade is designed to failover first to reduce downtime. The readers and write should be on the similar instance class in Aurora cluster. You can try either of these:
- Upgrade your new writer now, which would automatically failover to give your original configuration with both instances to new instance sizing.
- Manually failover the cluster to switch the configuration from console, CLI or API call, and then you can upgrade reader without a failover.
Note that when you perform a manual failover, you can specify which instance to fail over to, although specifying the instance ID is optional:
https://docs.aws.amazon.com/cli/latest/reference/rds/failover-db-cluster.html
What I like to do for Aurora clusters with >2 instances is put 2 of them in priority tier 0, and all the others in some other priority tier (2-15). That way, all the failovers default to switching between those 2 instances in the top priority tier and there's no worry about some other instance with a too-low capacity, wrong AZ, etc. being promoted to the writer.