First setup cloudbreak and install the cloudbreak CLI on your machine, don't forget to configure it to your cloudbreak deployer.
Run mpack.sh
to register the required mpacks with cloudbreak, this may show a failure on already registered mpacks, this is not a problem and can be ignored.
Run blueprint.sh
to upload the default blueprint for HCP.
Run recipes.sh
to upload the receipe scripts used.
Run python cluster.py --name NAME_YOUR_CLUSTER
to deploy an HCP cluster, then log in to cloudbreak and watch it go.
You can change the aws region and zone by adding aws: zone: us-east-2a region: us-east-2
also below values needs to be chaged based on the cloud provider
credentialName: cloudbreak-credential
hcpVersion: hcp version to be installed
cloudProvider: Cloud provider values can be (aws, ycloud, openstack)
yarnQueue: yarn queue if the cloud provider is 'ycloud' else neglect
to the config yaml.
You can force passwords and kerberos keys in config.yaml file. Any key set to RANDOM will get a 16 digit generated complex password.
template-xxx.j2 (aws or ycloud based on the cloud provider selected) is processed as a jinja template to create the cloudbreak template for each run. The temporary template actually used is deleted. You can change things like spot price, and instance types, mpack version to be installed in the template-xxx.j2 file if you know what you're doing with Cloudbreak.