Hi, I am writing since I have recently tried to deploy a 20-node cluster on Google Cloud. While I really liked the simplicity of Elasticluster workflow, I also noticed it took around 60 mins that to deploy such a cluster. For this reason, I was wondering whether it is possible to shorten the deployment time, by changing, e.g., some Ansible playbooks. At the moment, I am using the following configuration.
[cloud/google] provider=google # Configure Ansible SSH timeouts and the number of forked processes. [setup/ansible] ansible_forks=20 ansible_timeout=200 # Configure Slurm with 2 groups of hosts: 1 frontend (slurm_master) and # the rest with the role "slurm_worker". [setup/ansible-slurm] provider=ansible frontend_groups=slurm_master,r compute_groups=slurm_worker,r # Define the cluster. [cluster/myslurmcluster] cloud=google login=google setup=ansible-slurm security_group=default # Specify the image, machine type, node count, and boot disk size for # the cluster nodes. image_id=debian-10-buster-v20200910 flavor=n1-standard-4 frontend_nodes=1 compute_nodes=19 ssh_to=frontend boot_disk_size=50 # Configure the head node to have a 1 TB root volume using a # solid-state drive (SSD). [cluster/myslurmcluster/frontend] boot_disk_type=pd-ssd boot_disk_size=500 Thanks a lot for your help and your great work! Nicola -- You received this message because you are subscribed to the Google Groups "elasticluster" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticluster/910dd59d-c4cf-4519-8073-3e038ebf84bbn%40googlegroups.com.
