As Aurora 1 (Mysql 5.6) is reaching end of life soon, we tried to upgrade to Aurora 2 (mysql 5.7). We used a manual snapshot to create a new cluster. The upgrade went fine.
However as soon as we shifted our production load to this new cluster (2 instances of size r5.4xlarge), The overall performance of the application was hit badly. The reader instance was presenting a 99% CPU usage for whole time and many of our queries were taking many many minutes. That rendered everything unusable so we had to go back to 5.6 using the backup snapshot.
We used the cluster and instance parameters as close as possible between 5.6 and 5.7. I wonder what might be causing this and whats the way out?
I am in bit of a panic because we just have few weeks left.
AWS OFFICIALUpdated a year ago
