Configuring Static IPs in ParallelCluster for VPC Compute Fleet

AWS ParallelCluster does not natively support the use of static IPs for the compute fleet within a VPC through manipulation of the NetworkConfiguration section in the configuration file.

ParallelCluster uses Amazon EC2 Auto Scaling groups to manage the compute fleet, which dynamically scales instances up and down based on workload demands. These instances are typically assigned IP addresses dynamically from the subnet's CIDR range.

The networking configuration for ParallelCluster allows you to specify VPC settings, subnet configurations, and whether to use public IP addresses, but it does not provide direct control over assigning static private IP addresses to compute nodes.

You can configure aspects like the VPC ID, subnet IDs for head and compute nodes, and whether to use public IPs. However, there isn't a built-in mechanism to assign predetermined static private IPs to the dynamically scaled compute instances.

If static IP addressing is a critical requirement, you may need to explore alternative solutions or custom configurations outside of ParallelCluster's standard networking setup. This might involve additional networking layers or custom scripts, but would likely require significant customization beyond ParallelCluster's native capabilities.
Sources
Network configurations - AWS ParallelCluster
Configuration - AWS ParallelCluster
[vpc] section - AWS ParallelCluster