DNS Check
probe.dns resolves a DNS name and asserts the answers. The lookups run from the worker executing the flow, so the step verifies what the network around your automation sees: that a device's name resolves at all, that it resolves to the addresses you expect, or that a given resolver serves the right records. No device credentials are involved.
Using It
By default the step resolves each target's own name: every device from the step's target picker is looked up by its hostname (falling back to its management address). Set Name to Resolve to query one explicit name instead — the same question is then asked once per target. Direct-IP targets only make sense together with an explicit Name to Resolve; an IP literal is not a DNS question, and such a target fails with a clear message otherwise.
Record Type picks the query (A by default; AAAA, CNAME, MX, NS, and TXT are also available). Resolver optionally directs the query at a specific nameserver IP address; when empty the worker's system resolver is used. A resolver is a destination the step dials, so the flow's network policy must allow it on port 53 for both UDP and TCP (DNS falls back to TCP for large answers); host names, ports and URLs are refused. Expected Values lists answers, one per line, that must all appear in the response — comparison is normalized, so case, trailing dots, and surrounding quotes do not matter, and extra answers beyond the expected ones are fine. With no expected values, any successful resolution passes. Timeout (sec) bounds each attempt (default 5; steps saved with its old name, timeout_sec, keep working), and Attempts (default 1, up to 10) retries failed lookups with a half-second pause between tries. Max Parallel Devices sets how many targets are looked up at the same time (default 10, up to 32); 1 looks them up one after another. Evidence and metrics always follow target order.
Output
The summary reads dns <record type>: N/M lookups OK and is available to later steps as steps.NODE_ID.summary. Metrics record one <device name>_ms resolution latency per successful lookup. Each target adds one transport_meta evidence record named dns_<record type>_<device name> capturing the queried name, record type, resolver (system when none was set), the normalized answers, any missing expected values, and the error text when resolution failed.
Later steps read each target's result from steps.NODE_ID.output.targets['<device name>']: ok, error (empty when it passed) and latency_ms, plus the facts its evidence records (query_name, record_type, resolver, answers, missing_expected). A target that was never checked, for want of an address, still appears with ok false and its error. Inventory names are unique per site only, and an IP target is named by its address: a name two targets share is keyed as name (device id), so every result is kept.
When It Fails
The step fails when no target devices or IPs are configured, when Resolver is not an IP address, and when the flow's network policy does not allow the resolver (the policy's reason is then the step error) or the platform cannot apply the policy to it. A target fails when the lookup errors (NXDOMAIN, refusal, or timeout — the timeout applies per attempt, not to the step as a whole), when any Expected Value is missing from the answers, when the target has no name to resolve, or when it is a direct-IP target without an explicit Name to Resolve. One failing target fails the whole step, with the per-target reasons listed in the step error. Evidence is kept for every target that was actually looked up; a target skipped for having no name to resolve appears only in the step error.