Workflows Quick Start¶

This page assumes that you have a database and have completed the torc installation instructions. Refer to ArangoDB instance if you haven’t set up a database yet.

If you have not globally configured a connection to your database, refer to Database Connection.

This page runs a trivial workflow. Please refer to the [torc examples] (https://github.com/NREL/torc/tree/main/examples) or Tutorials for more complex workflows.

Create a workflow¶

Create a text file with one command per line. Here is one example file with dummy commands:

$ cat commands.txt
echo "Hello, World!"
echo "Hello, World!"
echo "Hello, World!"
echo "Hello, World!"
echo "Hello, World!"

Create the workflow.

$ torc workflows create-from-commands-file \
    --name my-workflow \
    --description "My workflow" \
    --cpus-per-job=1 \
    --memory-per-job=1g \
    --runtime-per-job=P0DT1H \
    commands.txt

2023-04-10 10:52:52,240 - INFO [torc.cli.workflows workflows.py:144] : Created a workflow from commands.txt with key=94956990

If you are in an HPC environment, add a Slurm scheduler configuration. If you are running locally, skip this step. Change account to your HPC account.
```
$ torc hpc slurm add-config -a account -w 01:00:00 -N debug-config -p debug
```

Start the workflow. This initializes the jobs, but does not run them yet.

$ torc -k 94956990 workflows start
2023-03-28 16:37:58,708 - INFO [torc.workflow_manager workflow_manager.py:99] : Started workflow

Start a worker to run the jobs.

If you are running locally, start a local worker.
```
$ torc -k 94956990 jobs run -p 1
```
If you are running in an HPC environment, start a worker with the Slurm scheduler. This will schedule one compute node with the Slurm scheduler defined in step 3.
```
$ torc -k 94956990 hpc slurm schedule-nodes -n 1
```

Monitor progress with torc or, if using an HPC with Slurm, squeue.

$ torc -k 94956990 jobs list

$ watch -n 10 squeue -u $USER

View the results when the workflow is complete.
```
$ torc -k 94956990 results list
```