Sui Node Monitoring

info

These instructions are for advanced users. If you just need a local development environment, you should instead follow the instructions in Create a Local Sui Network to create a local full node, validators, and faucet.

The examples below scrape localhost:9184/metrics, but the default full node template binds metrics to 0.0.0.0:9184. Restrict metrics-address with a firewall or bind it to a private interface in production.

You can view the metrics in the metrics UI, or use a tool like curl to get the metrics in a format that is easy to parse.

$ curl -s http://localhost:9184/metrics | grep -E 'sui_validator|sui_fullnode'

Key metrics

The following metrics are the most important for monitoring node health:

Metric	Type	What it tells you
highest_known_checkpoint	Gauge	Highest checkpoint the node knows about from the network
highest_synced_checkpoint	Gauge	Highest checkpoint fully synced (contents and effects)
last_executed_checkpoint	Gauge	Last checkpoint the node executed
last_executed_checkpoint_timestamp_ms	Gauge	Timestamp of the last executed checkpoint
current_epoch	Gauge	Current epoch ID
uptime	Counter	Node uptime in seconds (includes version and chain_identifier labels)
total_transaction_certificates	Counter	Total transaction certificates handled

Sync health check: Compare highest_known_checkpoint with highest_synced_checkpoint. If the gap grows, the node is falling behind. Compare your node's highest_synced_checkpoint against a trusted public endpoint to measure absolute lag:

# Check your node's sync progress
$ curl -s http://localhost:9184/metrics | grep -E 'highest_known_checkpoint|highest_synced_checkpoint|last_executed_checkpoint'

Production monitoring

For production monitoring, use Prometheus to scrape metrics and Grafana to visualize them.

Add a scrape target for your node in your Prometheus configuration:

scrape_configs:
  - job_name: 'sui-fullnode'
    static_configs:
      - targets: ['localhost:9184']
    scrape_interval: 15s

Suggested alerts

Set up alerts for the following conditions:

• Sync lag: highest_known_checkpoint - highest_synced_checkpoint > 100 sustained for 5 minutes indicates the node is falling behind.
• Node down: up{job="sui-fullnode"} == 0 sustained for 2 minutes indicates the node is unreachable.
• Execution lag: last_executed_checkpoint_timestamp_ms > 0 and time() - last_executed_checkpoint_timestamp_ms / 1000 > 120 sustained for 2 minutes indicates that the node is executing checkpoints more than 2 minutes behind chain time. This alert also fires while a healthy node catches up after an outage or snapshot restore. The > 0 guard avoids false alerts during startup, when the timestamp has not yet been populated. To detect a stalled executor separately, alert when last_executed_checkpoint does not advance over a suitable interval.