Reference — Metrics

The OPC UA adapter emits health and operational metrics through the EdgeCommons metric service. With metricEmission.target: messaging, metrics are published on the reserved UNS metric class:

ecv1/{device}/opcua-adapter/metric/{metricName}

The adapter never writes reserved metric topics directly. It defines metrics through MetricEmitter, so the same names, measures, units, and dimensions are used by messaging, CloudWatch, and Prometheus targets.

Dimension model

OPC UA metrics use a deliberately small CloudWatch dimension set. Each metric has the instance dimension, plus runtime-injected dimensions such as component identity and category. Operation and window semantics are encoded in measure names rather than high-cardinality dimensions.

Endpoint URLs, node ids, namespace URIs, browse names, and raw status text are not metric dimensions. Use data messages, events, logs, or command replies for those details.

`southbound_health`

The canonical southbound health metric: per-instance connection state, publish/poll latency, the read/stale/reconnect/write counters, and the subscribed-signal gauge.

Dimensions: instance.

Measure	Unit	Res	Meaning
`connectionState`	Count	1	`1` = OPC UA session up, `0` = down.
`publishLatencyMs`	Milliseconds	1	Last northbound publish round-trip (from the `data()` publish).
`pollLatencyMs`	Milliseconds	1	Last explicit-read round-trip (`repoll` / `sb/read`).
`readErrors`	Count	60	Read errors over the interval.
`staleSignals`	Count	60	Subscribed signals with no update for longer than `component.global.healthThresholds.staleSignalSecs` (default 30 seconds).
`reconnects`	Count	60	Session reconnects over the interval.
`writeErrors`	Count	60	`sb/write` entries that failed on the device path over the interval: the entry passed validation and the `writes.allow[]` allow-list and was then rejected by the server or aborted by an unavailable session. Allow-list refusals and caller errors (missing values, unresolvable refs, bad value types) are not counted — they surface in `OpcUaCommand.WriteFailure*`.
`signalsSubscribed`	Count	1	Gauge: the number of signals the instance’s session currently serves (the resolved monitored-item count); `0` while disconnected.

staleSignals counts each subscribed signal whose most recent update is older than component.global.healthThresholds.staleSignalSecs (default 30 seconds); see configuration reference.

`OpcUaCommand`

sb/* command activity: the whole command surface (every verb) plus explicit sb/read and sb/write detail.

Dimensions: instance.

Measure	Unit	Purpose
`CommandRequestTotal`	Count	Lifetime count of all `sb/*` command requests across every verb. Helps quantify command-plane demand.
`CommandRequestInterval`	Count	All `sb/*` command requests in the interval. Helps build current command-rate dashboards.
`CommandFailureTotal`	Count	Lifetime count of failed `sb/*` command requests across every verb. Helps identify persistent command-plane problems.
`CommandFailureInterval`	Count	Failed `sb/*` command requests in the interval. Helps alert on active command failures.
`ReadRequestTotal`	Count	Lifetime explicit `sb/read` samples returned. Helps quantify command-plane read demand.
`ReadRequestInterval`	Count	Explicit `sb/read` samples returned in the interval. Helps build current read-rate dashboards.
`ReadFailureTotal`	Count	Lifetime failed `sb/read` command requests. Helps identify persistent explicit-read problems.
`ReadFailureInterval`	Count	Failed `sb/read` command requests in the interval. Helps alert on active read failures.
`WriteRequestTotal`	Count	Lifetime `sb/write` entries issued to the server. Helps audit write activity.
`WriteRequestInterval`	Count	`sb/write` entries issued in the interval. Helps build current write-rate dashboards.
`WriteFailureTotal`	Count	Lifetime failed `sb/write` entries, including preflight rejection and bad server status. Helps audit failed control actions.
`WriteFailureInterval`	Count	Failed `sb/write` entries in the interval. Helps alert on active write failures.

`OpcUaSubscription`

Subscription data-change throughput and subscription shape.

Dimensions: instance.

Measure	Unit	Purpose
`SubscribedReadTotal`	Count	Lifetime subscription data-change samples received. Helps measure total streamed telemetry volume.
`SubscribedReadInterval`	Count	Subscription data-change samples received in the interval. Helps measure current telemetry rate.
`SubscriptionRecreateTotal`	Count	Lifetime subscription re-establishment attempts after transfer failure. Helps detect unstable sessions.
`SubscriptionRecreateInterval`	Count	Subscription re-establishment attempts in the interval. Helps alert on active subscription churn.
`SubscriptionCount`	Count	Active OPC UA subscription objects. Helps verify configured subscription grouping.
`MonitoredItemCount`	Count	Resolved monitored signal count. Helps confirm the live subscription inventory.

`OpcUaBrowse`

Address-space browse command use and response size.

Dimensions: instance.

Measure	Unit	Purpose
`BrowseRequestTotal`	Count	Lifetime `sb/browse` command requests. Helps quantify address-space exploration demand.
`BrowseRequestInterval`	Count	`sb/browse` requests in the interval. Helps spot active console or API browsing.
`BrowseFailureTotal`	Count	Lifetime failed `sb/browse` command requests. Helps identify browse permissions or server issues.
`BrowseFailureInterval`	Count	Failed `sb/browse` requests in the interval. Helps alert on active browse failures.
`BrowseReferenceTotal`	Count	Lifetime hierarchical references returned. Helps understand browse response volume.
`BrowseReferenceInterval`	Count	Hierarchical references returned in the interval. Helps detect expensive browse activity.
`BrowseTruncatedTotal`	Count	Lifetime browse responses truncated by `maxRefs`. Helps identify too-broad browse requests.
`BrowseTruncatedInterval`	Count	Truncated browse responses in the interval. Helps tune console/API browse depth and limits.

`OpcUaConnection`

Initial connection, terminal failures, and live session transitions.

Dimensions: instance.

Measure	Unit	Purpose
`ConnectionAttemptTotal`	Count	Lifetime initial OPC UA connection attempts. Helps detect startup retry loops.
`ConnectionAttemptInterval`	Count	Initial connection attempts in the interval. Helps alert on active retry storms.
`ConnectionFailureTotal`	Count	Lifetime failed initial connection attempts. Helps diagnose unavailable servers or bad endpoint config.
`ConnectionFailureInterval`	Count	Failed initial connection attempts in the interval. Helps alert on current connectivity failure.
`TerminalFailureTotal`	Count	Lifetime unretryable initial connection failures. Helps identify bad credentials, security policy, or certificate configuration.
`TerminalFailureInterval`	Count	Unretryable initial connection failures in the interval. Helps alert on configuration failures that will not self-heal.
`SessionDisconnectTotal`	Count	Lifetime active sessions becoming inactive after initial connection. Helps measure runtime instability.
`SessionDisconnectInterval`	Count	Active sessions becoming inactive in the interval. Helps alert on current server/network disruption.
`SessionReconnectTotal`	Count	Lifetime inactive sessions becoming active again. Helps confirm Milo session recovery.
`SessionReconnectInterval`	Count	Session reconnections in the interval. Helps detect churn even when the final state is connected.
`SessionConnected`	Count	`1` connected, `0` disconnected. Helps build current liveness dashboards.