# Test Launch of the Golem Cluster for Susteen

## Goals

The goal of the project is to create a Proof of Concept (PoC) for GPU machines in DataCenter.
Identify issues in the installation and configuration of such an environment.

## Plan
### Stage 1
Adding the following functionalities to the configurator and the automatic configuration file:

- **Multi-GPU Mode Selection**: Determine whether the machine should be exposed as a single machine with multiple GPUs or as several virtual machines, each with a single GPU.
- **Subnet and Payment Method Selection**: Allow for easier testing by isolating the environment through the choice of subnet and payment method.
- **Prometheus Server Selection**: Enable the selection of a Prometheus server to which node metrics will be pushed.

### Stage 2

- Create testing app and setup a server for testing a model. App shuld be based on the Worker API with an option to choose between an open network or a subnet.
- Delpoly grafana & prometeus to collect stats from cluster.

### Stage 3

- fixes and impovments in iunstaller for golem providers.
- solution for own requestors attestation.

# Project Timeline

- **21-02-2025:** First iteration of the test application
- **28-02-2025:** Setting up the susteen environment on Golem machines and delivering the first installation package
- **07-03-2025:** Grafana panel with basic information on the cluster's operation
- **03-03-2025 to 10-03-2025:** First iteration of tests
- **14-03-2025:** Releasing the second installation package based on reported issues
- **15-03-2025 to 28-03-2025:** Second iteration of tests
- **04-04-2025:** Final version of monitoring
- **18-04-2025:** Documenting conclusions and closing the PoC