HPC compute environments

Seqera Platform streamlines the deployment of Nextflow pipelines into both cloud-based and on-prem HPC clusters and supports compute environment creation for the following management and scheduling solutions:

• Altair PBS Pro
• Grid Engine
• IBM Spectrum LSF (Load Sharing Facility)
• Moab
• Slurm

Requirements

To launch pipelines into an HPC cluster from Seqera, the following requirements must be satisfied:

• The cluster should allow outbound connections to the Seqera web service.
• The cluster queue used to run the Nextflow head job must be able to submit cluster jobs.
• The Nextflow runtime version 21.02.0-edge (or later) must be installed on the cluster.

Credentials

Seqera requires SSH access to your HPC cluster to run pipelines. Use managed identities to enable granular access control and preserve individual cluster user identities.

You can also use workspace SSH credentials for cluster login, but this provides service account access to your HPC to all Platform users. This means that all users will be granted the same file system access, and all activity is logged under the same user account on your HPC cluster.

For HPC clusters that do not allow direct access through an SSH client, a secure connection can be authenticated with Tower Agent.

Seqera HPC compute environment

To create a new HPC compute environment:

1. In a Seqera workspace, select Compute environments > New environment.

2. Enter a descriptive name for this environment. Use only alphanumeric characters, dashes, and underscores.

3. Select your HPC environment from the Platform dropdown menu.

4. Select your existing managed identity, SSH, or Tower Agent credentials, or select + and SSH or Tower Agent to add new credentials.

5. Enter the absolute path of the Work directory to be used on the cluster.

   All managed identity users must be a part of the same Linux user group. The group must have access to the HPC compute environment work directory. Set group permissions for the work directory as follows (replace sharedgroupname and <WORKDIR> with your group name and work directory):

   chgrp -R sharedgroupname <WORKDIR>
   chmod -R g+wxs <WORKDIR>
   setfacl -Rdm g::rwX <WORKDIR>

   These commands change the group ownership of all files and directories in the work directory to sharedgroupname, ensure new files inherit the directory's group, and apply default ACL entries to allow the group read, write, and execute permissions for new files and directories. This setup facilitates shared access and consistent permissions management in the directory.

6. Enter the absolute path of the Launch directory to be used on the cluster. If omitted, it will be the same as the work directory.

7. Enter the Login hostname. This is usually the hostname or public IP address of the cluster's login node.

8. Enter the Head queue name. This is the default cluster queue to which the Nextflow job will be submitted.

9. Enter the Compute queue name. This is the default cluster queue to which the Nextflow job will submit tasks.

10. Expand Staging options to include:

    • Optional pre- or post-run Bash scripts that execute before or after the Nextflow pipeline execution in your environment.
    • Global Nextflow configuration settings for all pipeline runs launched with this compute environment. Configuration settings in this field override the same values in the pipeline Nextflow config file.

11. Specify custom Environment variables for the head job and/or compute jobs.

12. Configure any advanced options needed:

    • Use the Nextflow queue size to limit the number of jobs that Nextflow can submit to the scheduler at the same time.
    • Use the Head job submit options to add platform-specific submit options for the head job. You can optionally apply these options to compute jobs as well:

    Once set during compute environment creation, these options can't be overridden at pipeline launch time.

    In IBM LSF compute environments, use Unit for memory limits, Per job memory limits, and Per task reserve to control how memory is requested for Nextflow jobs.

13. Select Create to finalize the creation of the compute environment.

See Launch pipelines to start executing workflows in your HPC compute environment.