Recipes on how to provision Pulsar + HTCondor cluster in CESNET's MetaCloud. Recipes are written as generaly as possible but they are still quite specific to CESNET's MetaCloud and its OpenNebula deployment.
git clone https://github.com/Misenko/meta-galaxy-nodes-infra.git && cd meta-galaxy-nodes-infra- Create file
secrets.auto.tfvarsand populate it with OpenNebula and message queue credentials.
one = {
endpoint = "https://some.cloud.somewhere.cz:443/RPC2"
username = "user"
password = "password"
}
rabbitmq_galaxy_user_name = "galaxy-user"
rabbitmq_galaxy_user_password = "galaxy-user-password"
rabbitmq_admin_user_name = "admin-user"
rabbitmq_admin_user_password = "admin-user-password"
- Edit
variables.tfto your liking. - Unlock your SSH keyring, for example
ssh-add ~/.ssh/id_rsa. docker run -it --rm -v $(pwd):/app -v ${SSH_AUTH_SOCK}:/root/ssh -e "SSH_AUTH_SOCK=/root/ssh" misenko/terransibula apply
Last command will run a Docker container containing Terraform with OpenNebula provider and Ansible provisioner (https://github.com/Misenko/terransibula) so nothing special (except Docker) doesn't have to be installed on the host.
Recipes by default provision one NFS server node, one Pulsar + HTCondor master node and one HTCondor slave node. Number of HTCondor slave nodes can be configured in variables.tf.