Timestamp Change in RTP Header

[Kobster]

According to the RTP RFC (3550) Timestamp must increase in each RTP packets:
"The timestamp reflects the sampling instant of the first octet in the RTP data packet.  The sampling instant MUST be derived from a clock that increments monotonically and linearly in time to allow synchronization and jitter calculations ".

We are facing issues when the Timestamp  suddenly decreases (goes back to zero) in the middle of the call, without any telephony events (such as hold/transfer). the SSRC stays the same and Sequence number increments as expected. Only the Timestamp is acting up without clear explanation.

This behaviors is causing recording loss (these RTP packets are discarded by the Recorder that captures the RTP stream).

I would like to find out if decreasing the Timestamp means lack of adherence to the RTP RFC and also what would be the possible cause for such behavior ?

Thanks

Kobi 

[bijpraka]

What is the software version of the firmware on the device?

Thanks.

[Kobster]

This issue happens on Cisco 7942 phones with firmware Version: SCCP42.9-4-2SR3-1S.

On Cisco 8841 phones (and exact same environment) the issue is a bit different: when the timestamp resets the sequence number also resets, in the same time (packet), which does not lead to recording loss.

[Kobster]

Any ideas ? 

What if this phenomenon stems from live “bouncing” of the call across different media servers (similar to S4B’s adaptability capability) ?

This might happen due to latency/BW issues (or CPU/RAM overload on the media server) and it happens regardless of telephony events such as hold, transfer or conference.

 This issue happens only to the customer’s side so maybe this side is live-migrated to a more adjacent media server ?