Enabling ganglia with pcluster update

Environment:

• aws-parallelcluster-2.4.1
• OS: centos
• Scheduler: slurm
• Master instance type: t2.micro
• Compute instance type: c5.2xlarge

Bug description and how to reproduce: I tried enabling ganglia after first provisioning a ParallelCluster by adding extra_json = { "cluster" : { "ganglia_enabled" : "yes" } } to the config file. However, it seems that the chef recipe is not run following a pcluster update, as it is definitely ont installed now.

Is it possible to enable ganglia after first creating the MasterServer?

[joeydumont]

Ah, I see. Then my comment is only relevant when you manually edit /etc/chef/dna.json to enable ganglia on the MasterServer, because then you don’t go through the pcluster CLI.

[enrico-usai]

We changed the parameter from “cfncluster” to “cluster” when CfnCluster has been rebranded to AWS ParallelCluster, to remove the “cfncluster” word from the configuration file.

The code still supports the old “cfncluster” syntax, to maintain retrocompatibility with old configuration file, but we adapted the code to convert the new “cluster” syntax to “cfncluster” leaving the cookbook code untouched. See: https://github.com/aws/aws-parallelcluster/blob/v2.4.1/cli/pcluster/cfnconfig.py#L555

It means you can use “cluster” (new syntax) or “cfncluster” (old syntax) and both will be converted to “cfncluster” on the cookbook side.