OPERATORS / 02
Run with
visibility.
Keep build identity, network progress, disk growth, and recovery evidence close to the node.
Before deployment
Record the client commit, network, genesis configuration, data directory, and enabled RPC namespaces. Hardware sizing depends on the network, sync mode, retention, and workload; xOne does not publish a qualified universal minimum here.
Watch the complete system
Track the canonical head and lag, peer connectivity, import errors, process memory, disk growth, and restart behavior. Read the network-specific guide for the supported ports, metrics, and observation setup.
- Compare a recent block hash with a trusted independent endpoint.
- Separate process heap from file-backed memory when interpreting RSS.
- Keep enough free disk for database writes and recovery operations.
- Retain timestamps and build identity with each incident report.
Upgrade and rollback
Follow the version-specific compatibility notes. Preserve configuration and signer material, stop the node cleanly, and retain a recoverable copy before changing the binary or data format. Do not assume a newer database can be opened by an older binary.
Recovery begins with evidence
Use the documented recovery path for the selected storage mode. Repeatedly restarting or deleting state can obscure the original failure. Capture the first error and the last accepted block before taking action.
Reviewed 14 November 2026. Repository guides hold the current implementation details.
KEEP THE CURIOSITY GOING