You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For cost savings:
- Change the default AWS instance type we use for `hiveutil
create-cluster` and clusterpools from m6i.xlarge to m6a.xlarge. These
have the same specs, but we're told `i`(ntel) is more expensive than
`a`(md).
- Plumb a new `--aws-worker-instance-type` flag into `hiveutil
create-cluster` to allow customizing _just_ the worker instance type in
the generated install-config and machine pool. If unset, the previous
behavior will hold: we'll use `--aws-instance-type` for (masters and)
workers.
- Use the above to shrink the default worker instance type for e2e to
m6a.large, which has half the CPU and memory of m6a.xlarge.
HIVE-2426
flags.StringVar(&opt.AWSInstanceType, "aws-instance-type", clusterresource.AWSInstanceTypeDefault, "AWS cloud instance type for masters and workers (unless the latter is overridden by --aws-worker-instance-type)")
330
+
flags.StringVar(&opt.AWSWorkerInstanceType, "aws-worker-instance-type", "", "AWS cloud instance type for workers only. If unset, --aws-instance-type is used.")
329
331
330
332
// Azure flags
331
333
flags.StringVar(&opt.AzureBaseDomainResourceGroupName, "azure-base-domain-resource-group-name", "os4-common", "Resource group where the azure DNS zone for the base domain is found")
0 commit comments