Freeze one disputed transaction
Record the intended and observed Asterisk node, branch and maintenance release, manager session, exact action, Queue and Member filter presence, unique ActionID, send and receive times, raw frames, observed completion, collector version and the missing or disputed queue fact. Redact credentials and caller data.
Classify the first gap as wrong node, login failure, action unavailable, permission denial, malformed request, wrong queue or member key, response not correlated, series truncated, completion missing, parser loss, cache reuse or valid runtime state disputed by another source.
Prove target, session and action visibility
Confirm the request reached the intended Asterisk instance and that the manager account can see the selected action in the current connection. Reject target identity cached across failover, restart or reconnect.
Do not widen ACLs and manager classes while also changing the queue filter. Preserve the failed case, then make one reversible correction to the earliest demonstrated boundary.
Inspect filter semantics and exact identifiers
For QueueSummary, verify the optional queue filter. For QueueStatus, verify both optional queue and member filters and distinguish omission from an empty or wrongly normalized value. Compare queue names and member interfaces exactly as configured and returned.
A display name from a dashboard, endpoint label or agent login may not equal the member identifier expected by the manager action. An empty filtered result is not automatically an empty queue.
Require a correlated complete response series
Match every item and the completion boundary to both the current session and returned ActionID. Preserve raw frames and the installed release's event names rather than accepting only the first row the parser recognizes. Treat socket close, timeout, parse error, missing completion and a complete zero-item result as different outcomes.
If the collector pipelines queries, prove it cannot mix two queue snapshots by arrival order.
Compare the right independent evidence
Use effective queue configuration for configured membership and strategy, runtime device and member evidence for reachability or pause state, the queue log and channel records for caller history, and signaling or media evidence for a specific call. Compare the same node and time window.
A queue summary cannot explain a historical abandonment, and detailed status cannot prove that a phone rang, a person answered or audio passed.
Retest the narrow transaction and preserve limitations
Start a clean manager session, issue the intended filtered action with a fresh identifier, retain the complete series and compare it with one independent source that owns the disputed fact. Restore approved permission scope after diagnosis and document parser or cache repair.
Close only when the collector distinguishes complete, empty, denied, malformed and interrupted transactions. No PBX, queue, member, caller, manager configuration, socket, response series, completion marker, channel, packet, media stream or live call was inspected or tested for this diagnostic.
Do not diagnose a member change from an incomplete queue snapshot
The AMI queue-member diagnostic reconciles response, AGENT-class events and completed later state.
Do not mistake incomplete queue snapshots for a failed reload
The AMI queue reload or reset diagnostic checks exact scope, source ownership, generation and later overwrite first.
Do not diagnose caller control from a partial queue snapshot
The queue caller-control diagnostic requires complete framing before interpreting priority, position or absence.
Compare live AMI state with custom history carefully
The QueueLog diagnostic separates completed snapshots from custom event rows and report caches.
Sources & applicability
Primary references for the technical details above. Operational examples and planning checklists are VoIP.info editorial guidance.
Examples require adaptation to your topology. No live PBX or hardware testing is claimed.