Skip to main content
All metrics are served at GET /metrics on the metrics port and are exposed in the standard Prometheus text format. Configure the port using ports.metrics in the Helm chart.

Configuration

Metrics are configured in the node’s config.json:
Default histogram buckets for *_seconds metrics:

Build info

ton_node_build_info is an informational gauge that is always set to 1. Build metadata is encoded as labels:

Naming convention

All metrics follow the format:
  • ton_node_ – application prefix that prevents collisions in shared Prometheus instances.
  • {subsystem} – functional area defined in the table below.
  • {metric_name} – descriptive name in snake case.
  • {unit_suffix} – optional: _total (counters), _seconds (durations), _bytes (sizes), _ratio (dimensionless).

Metric types

Labels

Some metrics carry labels for additional dimensions:

engine

Core node state, including sync progress, masterchain tracking, validation intent, and applied transactions. This subsystem exposes 8 metrics: 7 gauges and 1 counter.

Sync status values

A healthy, synced node reports sync_status = 6. During initial sync or restart, the value progresses through 1 → 2 → 3 → 4 → 5 → 6.

Time difference

timediff_seconds measures the gap between wall-clock time and the gen_utime of the last applied masterchain block. Values below 20 seconds indicate a healthy node; values above 60 seconds require investigation.

validator

Block validation lifecycle, including state, set membership, outcomes, and gas. It exposes 8 metrics: 4 gauges, 3 counters, and 1 histogram.

Validation status values

collator

Block collation lifecycle, including outcomes, duration, gas usage, message flow, and transaction counts. It exposes 14 metrics: 2 gauges, 6 counters, and 6 histograms.

outqueue

Outbound message queue periodic cleanup statistics. This subsystem exposes 4 metrics, all of which are gauges.

ext_messages

External message queue containing messages received from clients and awaiting inclusion in blocks. It exposes 2 metrics: 1 gauge, 1 counter.

network

Networking metrics covering ADNL round-trips, catchain timings, overlay queries, and neighbor reliability. It exposes 7 metrics: 1 gauge, 1 counter, and 5 histograms.

db

Database operations, including shard state management, garbage collection, persistent state, and Merkle updates. It exposes 5 metrics: 1 gauge, 4 histograms.

block

Block parsing and size metrics. It exposes 3 metrics, all of which are histograms.