Hi all,
I would really appreciate some comments or first impressions about an issue I'm having; just having some lead from where to continue studying the problems would already be a lot of help:
I'm currently investigating an issue with some audio-only streams having constant audio drops in an otherwise idle system, with supposedly good network conditions. There are 2 local PCs (in the same LAN) that establish an audio call via a remote Kurento Media Server hosted in an Amazon AWS machine.
The audio has constant drops, and when these happen, the receiver Chrome instance prints 4 kinds of errors on console (timestamps removed, for clarity):
1: [WARNING:packet_buffer.cc(97)] Packet buffer flushed
2: [WARNING:rtcp_receiver.cc(399)] 210 RTCP blocks were skipped due to being malformed or of unrecognized/unsupported type, during the past 10 second period.
3: [WARNING:srtpsession.cc(135)] Failed to unprotect SRTP packet, err=9
4: [ERROR:srtptransport.cc(210)] Failed to unprotect RTP packet: size=1418, seqnum=10898, SSRC=436158332
KMS doesn't print any errors regarding RTCP, but it does some times complain about SRTP error code 9. So the RTCP errors show up only in Chrome, but the SRTP errors seem to be shared by both ends.
Normally, the audio drops coincide with bursts of messages 2, 3 and 4. Also, the '210' in messages of type 2 is a number that grows constantly. Over a 2 minutes call, this grew up to be around 900. I guess if I leave it for more time, this number will keep growing to the thousands.
I'd really be thankful for any comments about what kind of conditions are usually reasons for these warning and error messages. I've been doing all sort of tests:
- Enabling and disabling STUN / TURN
- Testing different machines
- Inspecting traffic with Wireshark. This one is interesting: NO invalid RTP or RTCP packets are found in Wireshark, while Chrome complains about hundreds of "malformed or unsupported" packets.
Kind regards,
Juan