Skip to main content

Node Lifecycle

The lifecycle of a Resultity node defines how it is initialized, connected, updated, and retired. Each stage is designed for automation, reliability, and minimal user intervention.

1. Installation and Initialization​

To become part of the Resultity network, a user installs the Node software on a compatible device. The installer:

  • Downloads the latest Agent and Runtime container images;
  • Generates a new EVM-compatible key pair (if not provided);
  • Prepares a local configuration and data directory;
  • Assigns a permanent node ID based on the generated key.

The node is now locally initialized but not yet connected to the network.

2. Registration and First Contact​

Upon first launch, the Agent:

  • Connects to the network endpoint (io.network.resultity.com);
  • Sends a registration request with node metadata and public key;
  • Receives an initial configuration snapshot;
  • Performs a test inference task to validate its readiness.

Once the test task is completed successfully, the node is marked as active and enters the scheduling pool.

3. Heartbeat and Monitoring​

To remain eligible for job assignment, a node must:

  • Send regular heartbeat pings with hardware status and model availability;
  • Respond to configuration checksums to ensure version alignment;
  • Accept administrative pings or soft shutdown commands from the network.

If a node misses a heartbeat window, it is temporarily excluded from job rotation until restored.

4. Configuration Sync​

Nodes periodically check the network for configuration changes. Each configuration bundle includes:

  • Approved models and required hashes;
  • Job execution parameters;
  • Runtime version mappings;
  • Participation rules for testnet and mainnet.

If a checksum mismatch is detected, the node automatically downloads and applies the updated configuration.

5. Updates and Maintenance​

The Agent handles self-updates for all core containers (Agent, Runtime, and approved models). Triggers for update include:

  • New configuration snapshots;
  • Expired model versions;
  • Runtime engine patches.

All updates are authenticated using network public keys and applied without manual intervention.

6. Shutdown and Restart​

When shut down, the node:

  • Closes all active WebSocket and HTTP sessions;
  • Caches incomplete job states (if applicable);
  • On restart, reconnects and re-validates the environment.

If the node was previously registered, no re-initialization is required.

The Resultity node lifecycle ensures stable, automated operation with minimal user management — while preserving traceability, integrity, and version consistency across the network.