Installation details
#ScyllaDB version: 5.4
#Cluster size: 18 nodes w/ ~4 TiB used.
os (RHEL/CentOS/Ubuntu/AWS AMI): Ubuntu
We are currently using scylla operator with kube and have built out a large cluster in a single availability zone. However, we want to move to having it in 3 AZs and wish to use racks to provide rack awareness to the zones to help mitigate any interzonal network costs.
However, we’ve hit issues with operator crashing or putting the entire data set on a single node when attempting this before. Are there any recommended ways of doing this so that we don’t create a major hot spot?