aws / aws-parallelcluster

AWS ParallelCluster is an AWS supported Open Source cluster management tool to deploy and manage HPC clusters in the AWS cloud.

Home Page:https://github.com/aws/aws-parallelcluster

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Add support for RHEL/Rocky 8.9

coderforlife opened this issue · comments

Required Info:

  • AWS ParallelCluster version: 3.8.0

Bug description and how to reproduce:
When creating a custom AMI based on RHEL or Rocky 8.9 it fails to build since it cannot access the repo at from https://fsx-lustre-client-repo.s3.amazonaws.com/el/8.9/x86_64/repodata/repomd.xml (only 8.8 exists). It is entirely possible that using the same exact repo would in fact work with 8.9, but that is not easy to change (at least I haven't found an easy way, but I don't really understand how to update Chef cookbooks in this system).

Another option could be to allow images to be built without FSx/Lustre support at all and skip all steps for that.

Hi, thanks for opening this feature request.

Currently the FSx-for-Lustre client does not support RHEL/Rocky 8.9 kernels (see the FSx-for-Lustre client kernel compatibility matrix).

We could consider a feature request to disable parts of the Chef cookbooks, but that's not something for the short term.

Alternatively, there's a way to use a custom cookbook to build an image. You can create a fork of the cookbook project and disable the installation of the lustre client.

I think it should be enough to comment out this line.

Would this approach satisfy you? If so, I can provide more information on how to use a custom cookbook at image build time.

That would be fantastic. There are a few other things I might want to change as well (such as disabling DCV which requires a massive amount of installs on Rocky because it has to install all of the Gnome desktop environment).

If you fork the cookbook and you disable the Lustre installation as described above (to be tested), you can use it in a development section of the image build configuration file:

Region: ...
Build:
  ...
Image:
  ...
DevSettings:
  Cookbook:
    ChefCookbook: <path_to_github_repo+branch>.tar.gz

Please make sure you do your changes on top of the release-3.8 branch on our repository.

Also, I'd say it's a good idea to use the same custom cookbook also at cluster creation, although in your very case it shouldn't matter. You can use the same DevSettings in your cluster configuration file.

Related to this: #5913

ParallelCluster 3.9.0 has been released with RHEL/Rocky 8.9 support.