Reference — Metrics
The UNS bridge emits relay metrics through the EdgeCommons metric service. With
metricEmission.target: messaging, metrics are published on the reserved UNS metric class:
ecv1/{device}/uns-bridge/main/metric/{metricName}The bridge originates its own metrics on the device bus and then relays them to the site broker like
other device traffic. It also forwards other components’ metric messages. This page describes only
the bridge’s own custom non-system metrics.
Dimension model
Section titled “Dimension model”The bridge’s metric measures encode UNS class or scalar counter names instead of adding per-class CloudWatch dimensions. Runtime-injected component dimensions identify the bridge component and device.
No topic, reply topic, endpoint, or error text is used as a metric dimension. Use logs and events for those high-cardinality diagnostics.
Emission model
Section titled “Emission model”Counters emit interval deltas. Gauges emit the current value. Deltas are saturating, so a process restart or counter reset never produces a negative value.
Per-class metrics use one measure per relayed class: state, cfg, evt, metric, data, log,
and app.
Per-class relay metrics
Section titled “Per-class relay metrics”Dimensions: runtime-injected component dimensions only.
| Metric | Measures | Unit | Purpose |
|---|---|---|---|
relay_uplinked |
state, cfg, evt, metric, data, log, app |
Count | Messages successfully relayed from device bus to site broker by class. Helps verify which traffic classes are crossing the bridge. |
relay_dropped_disabled |
state, cfg, evt, metric, data, log, app |
Count | Uplink messages dropped because the class is disabled by policy. Helps diagnose missing traffic caused by configuration. |
relay_dropped_rate |
state, cfg, evt, metric, data, log, app |
Count | Uplink messages dropped by token-bucket rate caps. Helps tune class-level rate limits. |
relay_dropped_disconnected |
state, cfg, evt, metric, data, log, app |
Count | Uplink messages dropped while the site link was down or after site publish failure when the class is not buffered. Helps quantify WAN outage loss. |
Scalar counter metrics
Section titled “Scalar counter metrics”Dimensions: runtime-injected component dimensions only.
Each metric uses one measure, count.
| Metric | Unit | Purpose |
|---|---|---|
relay_downlinked |
Count | Commands relayed from site to device. Helps measure remote control traffic. |
relay_loop_dropped |
Count | Messages dropped by hop-tag loop protection or max-hop checks. Helps detect relay loops or miswired bridges. |
relay_routed_dropped |
Count | Messages dropped by class-routing, device-pinning, or non-UNS-topic checks. Helps diagnose broker ACL or topic-shape problems. |
relay_malformed_dropped |
Count | Payloads that could not decode as valid EdgeCommons protobuf envelopes. Helps detect non-EdgeCommons publishers on relayed filters. |
relay_publish_failed |
Count | Forwarding decisions whose republish failed at the transport. Helps identify broker or network publish failures. |
relay_reply_relayed |
Count | Replies relayed from device back to site through the correlation map. Helps confirm request/reply proxying works. |
relay_reply_expired |
Count | Correlation entries removed without a reply because of TTL expiry or eviction. Helps detect slow or lost device replies. |
relay_reply_stray |
Count | Replies that arrived without a live correlation entry and were dropped. Helps identify late replies or duplicate reply traffic. |
relay_evt_buffered |
Count | Events buffered while the site link was down. Helps verify disconnect buffering is active. |
relay_evt_buffer_dropped |
Count | Events evicted from the full disconnect buffer. Helps size the event replay buffer. |
relay_evt_replayed |
Count | Buffered events replayed after site reconnect. Helps confirm recovery behavior. |
Gauge metrics
Section titled “Gauge metrics”Dimensions: runtime-injected component dimensions only.
| Metric | Measure | Unit | Purpose |
|---|---|---|---|
relay_pending_replies |
count |
Count | Current in-flight request/reply correlations. Helps detect buildup in proxied commands. |
site_connected |
connected |
None | 1 when the site broker connection is up, 0 when down. Helps drive bridge reachability alarms. |