MSP Workbench

📘

Source Code

You can find the full source code and documentation for this workflow on GitHub: MSP Workbench

MSP Workbench runs common MSP tasks across your tenants with one MSP credential, from a web UI or the command line. Every write goes through a read-only review and a single confirmation.

🚧

Note

This is a proof of concept built on PyCentral. Use it as a reference for your own MSP integrations, not for large-scale production use.

Each task has its own guide:

Prerequisites

• uv. The workflow requires Python 3.10+, which uv installs on demand.
• An HPE GreenLake MSP workspace with an API client credential.

Installation

1. Clone the repository and enter the workflow:

git clone -b v2 https://github.com/aruba/central-python-workflows.git
cd central-python-workflows/msp-workbench

2. Create the environment and install the dependencies:

uv venv --python 3.12
uv pip install -r requirements.txt

Credentials

MSP Workbench needs three values:

• Client ID and Client secret: a GreenLake API client credential created at the MSP workspace level.
• Workspace ID: the ID of the MSP workspace itself, not a tenant's.

👍

Tip

The MSP token exchange guide covers how to create an API credential and finding your MSP workspace ID.

You never enter a Central base URL; the cluster is found automatically.

In the web UI, enter them on the sign-in screen.

The command line reads them from token.yaml in the directory you run it from. Copy token.yaml.example to token.yaml and fill it in:

unified:
  client_id: <your-glp-client-id>
  client_secret: <your-glp-client-secret>
  workspace_id: <your-msp-workspace-id>  # the MSP workspace, not a tenant's

Start the Workbench

Start the server from the msp-workbench directory, then open http://127.0.0.1:8000/:

uv run python server.py

Demo Mode

Demo mode uses a fixed sample catalog, needs no credentials, and makes no live changes. The catalog has sample tenants, devices, and subscription keys that cover valid, insufficient-capacity, expired, not-yet-valid, and ineligible cases, plus Central data for monitoring.

• Web UI: choose Use demo mode on the sign-in screen.
• Command line: add --demo to any command.

Run a Job

Creating tenants, adding inventory, and assigning inventory all run the same way.

1. Review. Click Run preflight. This read-only check shows what will be written, what is already done, and what is blocked.
2. Confirm once. The confirm button, marked DEMO or LIVE WRITES, starts the run.
3. Watch the run. Progress updates as each step finishes. Leaving the page does not stop the run.
4. Stop, if needed. Stop after current step ends the run. You can't resume it; unprocessed items are marked skipped.

Writes go out in batches of five, and one job runs at a time. Jobs are kept in memory, so restarting the server loses them.

When the run ends, download its reports:

To retry, fix the cause, then import the failed-devices CSV in the task's CSV import panel. The new run goes through review and confirmation again. Work that already finished is re-checked, not repeated.

Common results when a run writes devices:

Command Line

The command line offers the same tasks, for example:

uv run python workbench.py --demo list tenants

list accepts tenants, services, devices, or subscriptions. These global flags go before the command:

Errors are printed as one JSON line on stderr with code, message, and hint. Exit codes:

Troubleshooting

Support

• Automation Team: [[email protected]](mailto:[email protected])
• Workflow Issues: GitHub Issues
• PyCentral Library: PyCentral Issues