fovus 1.0.6

The Fovus Python CLI

Getting Started with Fovus CLI

The official CLI tool for interacting with the Fovus Serverless HPC SaaS. Installation, setup, and usage documentation is provided here. Further detail is available via the CLI --help command or at our CLI documentation website.

Features include job creation (with validation), job status checking, and job file downloading. In addition, reusable job and user configurations with optional command-line overriding are supported. Additional features are detailed in this README and in our help documentation (linked above).

Installation

Prerequisites

1. Python and pip installation: https://www.python.org/downloads/
   1. Click the "Download Python" button.
   2. If you have an option of installing pip, select that option. (This should be the default.)
2. Install fovus using pip with the following command:

pip install fovus

Note: pip is usually bundled with Python. If you have Python and you don't have pip, you can install pip at the following link: https://pip.pypa.io/en/stable/installation/.

Initial setup

Before use, set up the following:

1. User config
2. Password environment variable

User config

A user config is for a pre-existing user that is already set up in the Fovus web UI.

The following values can be viewed on the Fovus web UI by clicking on your profile icon in the top right of the page:

• email: the email address used for your account.
• workspaceId: the ID for the workspace you would like to use. The workspace ID can be found by clicking on the user icon in the web UI.

Set up your CLI user config with the following command:

fovus --configure

Note:

1. This user config will save to ~/.fovus/user_configs/FOVUS_default_user_config.json.
2. To view all your configs, use fovus --open-config-folder.
3. The user config can also be provided in the following ways (these will override default user config):
   1. Using the following parameters: --email your_email and --workspace-id your_workspace_id.
   2. Using the following parameter: --user-config-file-path /path/to/your/user_config.json

Password environment variable

Export your Fovus password as an environment variable by following the instructions for your system:

Unix-based systems

export FOVUS_CLI_PASSWORD='your_password_here'

Note: if you are using a Unix-based system and your password contains a single quote ', replace all instances of the single quote ' with the following character sequence: '\''.

Windows

1. Run setx FOVUS_CLI_PASSWORD "your_password_here" in command prompt
2. Close your command prompt and open a new one (to refresh the environment variables)

Post-installation setup

Run fovus --configure and follow the prompts. If this results in an error, see below for common errors

Windows - Python can't find script

1. Run echo %USERPROFILE%\AppData\Roaming\Python\PythonXX\Scripts, replacing PythonXX with your version (can find using python--version). For example, if your python version is 3.11.2, replace PythonXX with Python311
2. Run setx PATH "%PATH%;C:\path\to\scripts", replacing C:\path\to\scripts with the modified output from step 1.
3. Close and reopen your command prompt.

Basic usage

See the "Advanced Usage" section for additional options regarding file inclusion/exclusion for create job and download job files.

Viewing your config files

fovus --open-config-folder

Note:

1. Contains empty job and user config files for reference and copying.

Creating jobs

This command will upload files from the specified directory to Fovus and create a job.

Upon job creation:

1. The following will be validated: your job configuration structure (if provided as JSON), your job configuration values against your chosen benchmarking profile, and your software configurations (if job is monolithic).
2. A .fovus folder will be created in your local file root directory, which will contain a job_data.json file (i.e., /path/to/job/file/root/directory/.fovus/job_data.json. This json is used to facilitate checking job status and downloading job files by tracking the job's jobID.

fovus --create-job --job-config-file-path "/path/to/job/config/file.json" --job-file-root-directory "/path/to/job/file/root/directory"

Note:

1. To create your own job template, do the following:
   1. Open the fovus config folder with --open-config-folder.
   2. Make a copy of either ~/.fovus/job_configs/FOVUS_job_template_containerized.json or ~/.fovus/job_configs FOVUS_job_template_monolithic.json.
   3. IMPORTANT: rename your copy, or it will be overwritten.
   4. Fill in the fields.
2. Each subdirectory in your job file root directory will be treated as its own task.
   1. Spaces are not allowed in task directory names.
3. You can add the optional argument --job-name "your_job_name_here" to set a custom job name.

Getting job status

If you created the job using the Fovus CLI on the same local machine (i.e., the .fovus folder was created in your job root directory -- see "Creating job" section for details):

fovus --get-job-current-status --job-file-root-directory "/path/to/job/file/root/directory

If you did not create the job using the Fovus CLI:

fovus --get-job-current-status --job-id "your_job_id"

Note:

1. Job ID is viewable in the Web UI.

Downloading completed job files

If you created the job using the Fovus CLI on the same local machine (i.e., the .fovus folder was created in your job root directory -- see "Creating job" section for details):

fovus --download-job-files --job-file-root-directory "/path/to/job/file/root/directory"

If you did not create the job using the Fovus CLI:

fovus --download-job-files --job-file-root-directory "/path/to/job/file/root/directory" --job-id "your_job_id"

Note:

1. Only new or updated files will be downloaded (checked using file hash).

Advanced usage

Include/Exclude file patterns

When creating a job, the arguments --include-input and --exclude-input are available to you. These arguments are used to include/exclude filepaths for upload/download. Valid values for these arguments are alphanumeric along with the wildcards: * (matches any number of characters), ? (matches any single character).

For example:

• --include-input "Car*/*" "config?/*" includes:
  • All files under any directory that begins with Car.
  • All files in folders titled configX, where X is any character.
• --exclude-input "Car*/*.log" "local?/*" uploads all files except:
  • All .log filetypes within every directory that begins with Car.
  • All files in any directory titled localX where X is any character.
• The same logic applies for --include-output and --exclude-output except those are used for downloading job files.

Usage with create job (--create-job)

1. --include-input will only upload filepaths (relative to the job root directory) that match the expressions provided.
2. --exclude-input will upload all filepaths except those that match the expression provided.

Usage with download job files (--download-job-files)

1. --include-output will only download filepaths (relative to the job root directory) that match the expressions provided.
2. --exclude-output will download all filepaths except those that match the expression provided.

Overriding

Any value in a job config or a user config can be overriden by providing the respective json key as an argument. Examples are given below, and more details are available in --help.

Single-value arguments

All single-value arguments (i.e., not lists) can be overriden using a single argument and a value. For example, the benchmarkingProfileName in your job config, you can do so by providing the argument --benchmarkingProfileName "your_benchmarking_profile_name_here".

Lists of values

All lists of values can be overridden using a single argument and a list of values. For example, if you want to override the supportedCpuArchitectures in your job config, you can do so by providing the argument --supportedCpuArchitectures "x86_64" "arm64".

Lists of objects

Currently, the only list of objects that exists is the monolithicList, which is only used for monolithic jobs. The only value that can be overridden in a monolithicList is licenseCountPerTask. To do this, use the command --monolithic-override "vendor_name" "software_name" "license_feature" "your_new_license_count_here".

Note:

1. The monolithicList object must already exist in the job config JSON.
2. The vendor name, software name, and license feature are used to reference the specific object within the list and do not override existing values.
3. If an object that does not match the given vendor name, software name, and license feature does not exist, an error will be thrown.