centerforaisafety / cerberus-cluster

HPC cluster code and configurations for running on OCI
Universal Permissive License v1.0
4 stars 0 forks source link

added configure_for_weka.sh script #228

Closed andriy-safe-ai closed 1 year ago

andriy-safe-ai commented 1 year ago

Added script to configure a new node to support weka. Updated the existing script to take the path to the host file as a command line argument. Added description to help explain what this script does. Added error checking.

steven-safeai commented 1 year ago

Should describe the format of the hosts file. Would be nice if it can just read the Ansible hosts file.

andriy-safe-ai commented 1 year ago

I like the idea of specifying the format for the hosts file in the description. I'll add that. Previously the script just had a hardcoded path to the hosts file so I thought that specifying a path was an improvement. We don't need to run this script on node a second time but I could make sure it is idempotent so we don't need to worry about filtering the nodes. That why each time we run this script it will run on all of our compute nodes each time. At the moment, we would only add the hosts we are trying to add to the cluster and not any other node. I could also go further and convert this script into an Ansible playbook however that might be better to do once I'm working on adding Weka to our Ansible playbooks. With this is just wanted to make sure that the script is in source control and that it is easier and safer to work with.

steven-safeai commented 1 year ago

I'm okay skipping the reference to Ansible hosts file but should describe the host file format at least in here in case there's an error or at least to remove confusion.

andriy-safe-ai commented 1 year ago

Squashed and merged.