Diagnose Asterisk calls that disconnect
Record the time from call start and from answer to disconnection. A repeatable interval is a clue, not a diagnosis. Find which participant or application ended the call and what happened immediately beforehand.
Source-reviewed, 19 October 2026. Prepared from the cited documentation. No disconnection scenario has been independently tested by Phone Guides. Scope: Asterisk 22/PJSIP; timer and carrier behaviour require the exact deployed references.
Establish the stage and affected leg
Separate a call attempt rejected before answer from an answered conversation that ends unexpectedly. Include whether the failure follows hold, transfer, a menu timeout or a particular destination. Record recent configuration and network changes.
Use first troubleshooting checks to obtain a bounded reproduction. Correlate the incoming and outgoing channels without assuming that a multi-leg call has only one final cause.
Read termination with context
| Observation | Investigation |
|---|---|
| Remote party sends termination | Determine which peer and what its own evidence reports |
| Asterisk ends a dialplan path | Inspect application completion, timeout and subsequent priorities |
| Transaction repeatedly retries | Determine which response or acknowledgement is missing |
| Session refresh fails | Compare negotiated refresh behaviour and later requests/responses |
| Media stops before termination | Inspect RTP reachability and any configured media timeout |
| Service restarts or crashes | Inspect process, operating-system and restart evidence |
The PJSIP reference documents session-timer and RTP-timeout options. Read their exact definitions, defaults and scope before changing them. Increasing several timers together can hide the original sequence.
Check application behaviour
An intended dial timeout, queue exit, IVR timeout or integration fallback can sound like a network drop. Inspect the loaded route and application result at the event time. Compare the Dial application reference with the actual arguments.
For integrated applications, include dependency timeouts, ARI resource cleanup and caller hangup handling. A late asynchronous response must not be treated as permission to create a replacement call automatically.
Preserve evidence from both ends
Collect timestamped signalling, relevant application messages and process events. Request the carrier or endpoint's matching evidence through the appropriate private support path. A local cause value can describe what Asterisk observed without proving the remote system's underlying reason.
Use CDR/CEL and logs for their different reporting scopes. Avoid treating one final CDR row as a complete event sequence. Redact subscriber details and credentials before sharing extracts.
Correlate application and channel results
After Dial() returns, DIALSTATUS describes that Dial attempt. Values such as CHANUNAVAIL, CONGESTION, NOANSWER, BUSY and ANSWER narrow the stage; ANSWER means the call was answered, not that a later disconnect was normal. Where multiple outbound channels were tried, the HANGUPCAUSE functions can expose per-channel cause details after Dial completes. Preserve both the technology cause and Asterisk's translated cause when available.
For event collection, the Asterisk 22 AMI Hangup event is emitted per channel and carries Uniqueid, Linkedid, Cause and Cause-txt; TechCause is present for off-nominal termination. Correlate the related channel IDs and preceding requests rather than assigning the whole call to one hangup event. CEL records can add channel and bridge event order. A CDR remains a reporting record, not a full signalling transcript. A cause code records Asterisk's view of the channel result and may not establish the remote system's underlying reason.
Confirm a correction
Change one supported value and repeat the affected route for longer than its previous failure interval. Also check a normal hangup by each party and the relevant hold/transfer path. Confirm that channels and application resources clear afterwards.
Retain the previous configuration and document whether the cause was established or only narrowed. If audio stopped first, follow one-way and missing audio as a separate stage.
Return to Asterisk documentation.