Freeze one call segment and the disputed value
Record timestamp and timezone, direction, channel names and unique IDs, linked ID, endpoints, bridge transitions, media type, codec, SSRC, the exact CHANNEL(rtcp,...) expression, returned value and collection point. Preserve the user symptom in directional terms: who heard choppiness, silence, delay, echo or clean audio.
Classify the first mismatch: empty result, zero count, missing remote value, changing SSRC, implausible loss or jitter, score inconsistent with counters, good score with bad audio, bad score with acceptable audio or only one call leg collected.
Prove the function ran on the intended channel
Confirm that the evaluated owner is a live PJSIP channel supporting the documented item and that the requested audio or video stream exists. Follow Local-channel halves, transfers, queues and created outbound channels explicitly; shared linked IDs do not merge their RTCP state.
Collect the value while the intended owner remains available. If a hangup handler is used, retain which channel owns it and whether later work copied the result. Do not turn an unavailable item into numeric zero.
Separate Asterisk observations from peer RTCP reports
Identify whether the disputed field comes from Asterisk analysis of received RTP or from sender and receiver reports supplied by the remote endpoint. Verify whether RTCP packets exist, their source, report timing, SSRC and the interval they cover.
A missing peer report cannot be repaired by relabeling a local statistic. A remote implementation may calculate report fields differently, and a late report may describe a different portion of the call than the user complaint.
Reconstruct topology and direction changes
Preserve bridge technology, direct-media transitions, hold, transfer, re-INVITE or UPDATE exchanges and SSRC changes. Determine when Asterisk was in the packet path for each channel and whether the measured stream still carried the conversation.
If media bypassed Asterisk, use endpoint and network evidence for that interval rather than treating a PBX-side gap as zero impairment. If the call gained a new channel or stream, retain both segments instead of overwriting the earlier identity.
Compare raw counters before interpreting a score
Review packet counts, lost-packet values, jitter and round-trip time from the same channel, direction and interval before interpreting MES. Keep local and reported summaries separate and preserve the units and field names exported by the selected interface.
MES is a network-derived opinion from RTT, jitter and loss, not a listening score. Clean counters cannot exclude clipping, echo, device gain, acoustic problems, wrong source selection or unintelligible content, while poor counters do not locate the failing queue or link by themselves.
Retest one boundary and preserve independent evidence
Correct one demonstrated collection, reporting, topology or network boundary. Repeat the same authorized call pattern and collect the intended channel stats, bounded packet sample, endpoint counters and directional listening result on aligned timestamps. Confirm a nearby unaffected path so the change was not a blanket workaround.
Remove temporary debug, restrict sensitive captures and document retention. No PBX, channel, bridge, RTP or RTCP stream, endpoint, packet, network path, call or audible result was inspected or tested for this diagnostic.
Sources & applicability
Primary references for the technical details above. Operational examples and planning checklists are VoIP.info editorial guidance.
- Asterisk 22 CHANNEL function ↗
- Asterisk Media Experience Score ↗
- Asterisk MES implementation ↗
- Asterisk bridge media paths ↗
- Asterisk PJSIP troubleshooting: direct media ↗
Examples require adaptation to your topology. No live PBX or hardware testing is claimed.