Run Profiles

A run profile says where a workflow's steps run: on this computer, on a SLURM cluster or on AWS Batch. It also sets the container engine and default resources. Profiles are kept in Global Settings, for every workflow.

Making a profile

Open Global Settings on the home page and, under Run profiles, click SLURM cluster profile, AWS Batch profile or Local profile. Click Save global settings when done. Each starts from sensible values; change them in its card:

Setting	Meaning
Name	The Nextflow profile name, used as -profile <name>. Letters, digits and _; not docker, which every project already has.
Runs steps on	This computer, SLURM cluster or AWS Batch.
Partition (queue) / Job queue	The SLURM partition, or the AWS Batch job queue. Empty means the cluster's default partition.
Containers	Docker, Singularity, Apptainer, or none (the tools must be installed where steps run). Clusters usually offer Singularity or Apptainer.
CPUs, memory, time per step	Defaults for steps that don't set their own. On SLURM, the time must fit the partition's limit.
Jobs at once	How many jobs Nextflow keeps queued.
Extra sbatch options	SLURM only, for example --account=lab.
AWS region, S3 work directory	AWS Batch only, for example eu-west-1 and s3://my-bucket/work.

Problems with a profile show in its card; a profile with problems is left out of exports and can't be chosen for runs.

Using a profile

• In an exported project (every kind of profile): every profile goes into its nextflow.config, and its README lists them. Run with nextflow run main.nf -profile slurm. A project with an nf-core Pipeline node gets them in profiles.config instead: nextflow run nf-core/<pipeline> … -profile test,slurm -c profiles.config. See Exporting projects.
• For runs from N-WAVE (local and SLURM profiles): choose the profile all workflows use under Runs use by default in Global Settings. For a single workflow, turn off Use the global run profile in its Execution Settings → Profiles and choose another. The Run button's tooltip and the run's first lines name it. This computer uses the other execution settings.

Workflows made before profiles moved to Global Settings may still keep profiles of their own; they can still be used, and Execution Settings → Profiles offers Move to Global Settings.

Running on SLURM from N-WAVE

N-WAVE submits each step with sbatch, so its backend has to run where jobs can be submitted, such as a login node:

• Run N-WAVE from source with Nextflow installed and NEXTFLOW_EXECUTION_MODE=local, with sbatch on the PATH.
• Keep the output directory on storage the compute nodes share: each step runs as a job in the run's nextflow/work folder.

The Docker installation runs Nextflow in a container that can't reach a scheduler; a SLURM run there stops and says so. Export the project and run it on the cluster instead.

Cancel also cancels the run's SLURM jobs, and the run panel shows the steps as usual. The server's resource limits still cap each step.

AWS Batch profiles are for exported projects only: N-WAVE doesn't run them itself yet. A run needs AWS credentials where Nextflow runs, and the AWS CLI in the queue's compute environment (see Nextflow's AWS Batch guide).

See also

• Troubleshooting: SLURM jobs wait forever
• Nextflow executors and container engines