17 thoughts on - Asterisk Behind NAT Early Media Video

• You would need to examine things specifically and see where media is flowing. Is the recipient behind NAT? If so then until we receive media from them (wich may or may not occur with early media) we may not have the correct target of media.

  Cheers,

• Yes, media is flowing through Asterisk because both client’s are behind different NAT’s.

  Do I need to do something special in the Call Flow? Or anything additional to the pjsip.conf?

  2018-04-09 16:50 GMT+02:00 Joshua Colp :

• This doesn’t answer the question of what is ACTUALLY happening in the scenario you describe which is very important.

  The “rtp_symmetric” option as you’ve used causes Asterisk to send media to the source of media, but it requires us to receive media. If we don’t receive it then we send media to where they’ve told us to send it, which as I’ve mentioned can be wrong.

• My understanding based on Wireshark analysis is that the signaling works
  (also the recipent phone is displaying the video frame before accepting the call), also the calling phone send video (i see that also via Wireshark)
  but the recipent phone doesn’t get any video from the Asterisk before the call.

  2018-04-09 17:02 GMT+02:00 Joshua Colp :

• Ah yeah video, I forgot that it was a recent change to add support for it[1]. It’s not yet in any release.

  [1] https://gerrit.asterisk.org/#/c/8398/

• wohoo, so if I unterstand it correctly with that patch early media video works over the Asterisk server? In other words the Asterisk server get’s able to (process/)forward the early media video stream with that patch?

  2018-04-09 17:57 GMT+02:00 Joshua Colp :

• The patch forwards video while in an early media state before the call is answered and bridged, yes.

• Hi Benjamin!

  You’re obviously using a similar scenario that I have in place for testing. I initially had issues with early media (not only video also audio) as well in that scenario. What I had to do was to additionally set

  external_media_address=

  in pjsip.conf

  Also, as I wrote the patch for early-media video I’d be interested in any feedback from it.

• I applied the patch to my Asterisk 13.20. But it seems that it still doesn’t forward the early media video stream. Do I need to put something special into the extensions.conf? I basically just make a Dial. The calling Client sends the 183 protocol.

  [public]
  exten => 6001,1,Dial(SIP/${EXTEN})

  2018-04-09 18:14 GMT+02:00 Joshua Colp :

• Hi Florian

  I already have the external_media_address set in the PJSIP setup. Also the external_signaling_address is set to the Public IP. If I make a call from an Early Media (video&audio) capable device to an Early Media capable device (also video&audio) the Early Media audio works perfectly. But no video. If I sniff with wireshark on the recipent device I just see G711
  (audio) RTP traffic. The h264 RTP traffic is missing before I accept the call. After accepting the call the h264 RTP traffic comes through.

  The 183 SIP protocoll comes through. Even Asterisk is noticing it:
  — PJSIP/6002-00000013 is making progress passing it to PJSIP/6001-00000012

  I tried both Asterisk 15 with pjsip.conf configuration and Asterisk 13 with sip.conf (chan_sip). In both cases I just put the both case AST_FRAME_VIDEO: statements before the two voice cases, like in your diff and recompiled/reinstalled.

  Regards

  Benjamin

  2018-04-10 9:37 GMT+02:00 Floimair Florian :

• I just noticed, the calling device isn’t even sending the early media video stream. It just sends an early media audio stream. Is there propably a change in the signaling needed?

  (On another P2P SIP Server the early media video works.)

  2018-04-10 12:29 GMT+02:00 Benjamin Marty :

• I think I found the root cause. The H264 Early Media video is received successfully on the Asterisk Server. It also seems to get processed. But it’s send to the private IP of the receipent SIP phone.

  For clarification:
  178.82.XX.XX is my Public IP of my Internet access. Both phones use this as Public IP via standard Source NAT.
  159.89.XX.XX is the IP of the Asterisk Server. For this test I used a Server without Destination NAT. So the eth0 interface has this IP.

  Packet capture:
  No. Time Source Destination Protocol Length Info
  141 2018-04-11 06:40:03.306561 178.82.XX.XX 159.89.XX.XX
  H264 64 PT=H264, SSRC=0x3CB1E12D, Seq=19561, Time=319121408
  SPS

  Frame 141: 64 bytes on wire (512 bits), 64 bytes captured (512 bits)
  Ethernet II, Src: JuniperN_34:67:f0 (40:a6:77:34:67:f0), Dst:
  da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7)
  Internet Protocol Version 4, Src: 178.82.169.0, Dst: 159.89.104.193
  User Datagram Protocol, Src Port: 5006, Dst Port: 13182
  Real-Time Transport Protocol H.264

  No. Time Source Destination Protocol Length Info
  142 2018-04-11 06:40:03.306682 159.89.XX.XX
  192.168.XX.XX H264 64 PT=H264, SSRC=0x5EE97C55, Seq=30572, Time=319121408 SPS

  Frame 142: 64 bytes on wire (512 bits), 64 bytes captured (512 bits)
  Ethernet II, Src: da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7), Dst:
  IETF-VRRP-VRID_6e (00:00:5e:00:01:6e)
  Internet Protocol Version 4, Src: 159.89.104.193, Dst: 192.168.1.185
  User Datagram Protocol, Src Port: 10298, Dst Port: 5022
  Real-Time Transport Protocol H.264

  PJSIP.conf:
  [7004]
  type = endpoint context = internal rewrite_contact = yes direct_media = no rtp_symmetric = yes
  ;force_rport = yes disallow = all allow = g722, alaw, ulaw, gsm, ilbc, h264
  aors = 7004
  auth = auth7004

  [7004]
  type = aor max_contacts = 2

  [auth7004]
  type=auth auth_type=userpass password=1234
  username=7004

  extensions.conf:
  [internal]
  exten => _700X,1,Dial(PJSIP/${EXTEN})

  2018-04-10 16:43 GMT+02:00 Benjamin Marty :

• SSBkaWQgYSBxdWljayBjaGVjayBiZXR3ZWVuIHdoYXQgSSBoYXZlIHNldCBhbmQgeW91ciBzZXR0
  aW5ncyBiZWxvdy4NCg0KWW91IGNhbiB0cnkgdGhlIGZvbGxvd2luZyBhbmQgc2VlIGlmIGl0IGhl bHBzDQoNCkluIHlvdXIgZW5kcG9pbnQ6DQpiaW5kX3J0cF90b19tZWRpYV9hZGRyZXNzPXllcw0K
  DQoNCg0KDQpXaXRoIGJlc3QgcmVnYXJkcw0KDQpGbG9yaWFuIEZsb2ltYWlyDQpJbm5vdmF0aW9u IC0gU29mdHdhcmUtRGV2ZWxvcG1lbnQgLSAgVm9JUCAmIERldk9wcw0KDQpDT01NRU5EIElOVEVS
  TkFUSU9OQUwgR01CSA0KQS01MDIwIFNhbHpidXJnLCBTYWFsYWNoc3RyYcOfZSA1MQ0KVGVsOiAr NDMtNjYyLTg1IDYyIDI1DQpGYXg6ICs0My02NjItODUgNjIgMjYNCmh0dHA6Ly93d3cuY29tbWVu ZC5jb20NCg0KU2VjdXJpdHkgYW5kIENvbW11bmljYXRpb24gYnkgQ29tbWVuZA0KDQpGTiAxNzg2
  MTh6IHwgTEcgU2FsemJ1cmcNCg0KVm9uOiBhc3Rlcmlzay11c2Vycy1ib3VuY2VzQGxpc3RzLmRp Z2l1bS5jb20gW21haWx0bzphc3Rlcmlzay11c2Vycy1ib3VuY2VzQGxpc3RzLmRpZ2l1bS5jb21d IEltIEF1ZnRyYWcgdm9uIEJlbmphbWluIE1hcnR5DQpHZXNlbmRldDogTWl0dHdvY2gsIDExLiBB
  cHJpbCAyMDE4IDA4OjU1DQpBbjogQXN0ZXJpc2sgVXNlcnMgTWFpbGluZyBMaXN0IC0gTm9uLUNv bW1lcmNpYWwgRGlzY3Vzc2lvbiA8YXN0ZXJpc2stdXNlcnNAbGlzdHMuZGlnaXVtLmNvbT4NCkJl dHJlZmY6IFJlOiBbYXN0ZXJpc2stdXNlcnNdIEFzdGVyaXNrIGJlaGluZCBOQVQgRWFybHkgTWVk aWEgVmlkZW8NCg0KSSB0aGluayBJIGZvdW5kIHRoZSByb290IGNhdXNlLiBUaGUgSDI2NCBFYXJs eSBNZWRpYSB2aWRlbyBpcyByZWNlaXZlZCBzdWNjZXNzZnVsbHkgb24gdGhlIEFzdGVyaXNrIFNl cnZlci4gSXQgYWxzbyBzZWVtcyB0byBnZXQgcHJvY2Vzc2VkLiBCdXQgaXQncyBzZW5kIHRvIHRo ZSBwcml2YXRlIElQIG9mIHRoZSByZWNlaXBlbnQgU0lQIHBob25lLg0KRm9yIGNsYXJpZmljYXRp b246DQoxNzguODIuWFguWFggaXMgbXkgUHVibGljIElQIG9mIG15IEludGVybmV0IGFjY2Vzcy4g Qm90aCBwaG9uZXMgdXNlIHRoaXMgYXMgUHVibGljIElQIHZpYSBzdGFuZGFyZCBTb3VyY2UgTkFU
  Lg0KMTU5Ljg5LlhYLlhYIGlzIHRoZSBJUCBvZiB0aGUgQXN0ZXJpc2sgU2VydmVyLiBGb3IgdGhp cyB0ZXN0IEkgdXNlZCBhIFNlcnZlciB3aXRob3V0IERlc3RpbmF0aW9uIE5BVC4gU28gdGhlIGV0
  aDAgaW50ZXJmYWNlIGhhcyB0aGlzIElQLg0KUGFja2V0IGNhcHR1cmU6DQpOby4gICAgIFRpbWUg ICAgICAgICAgICAgICAgICAgICAgICAgIFNvdXJjZSAgICAgICAgICAgICAgICBEZXN0aW5hdGlv biAgICAgICAgICAgUHJvdG9jb2wgTGVuZ3RoIEluZm8NCiAgICAxNDEgMjAxOC0wNC0xMSAwNjo0
  MDowMy4zMDY1NjEgICAgMTc4LjgyLlhYLlhYICAgICAgICAgIDE1OS44OS5YWC5YWCAgICAgICAg SDI2NCAgICAgNjQgICAgIFBUPUgyNjQsIFNTUkM9MHgzQ0IxRTEyRCwgU2VxPTE5NTYxLCBUaW1l PTMxOTEyMTQwOCBTUFMNCg0KRnJhbWUgMTQxOiA2NCBieXRlcyBvbiB3aXJlICg1MTIgYml0cyks IDY0IGJ5dGVzIGNhcHR1cmVkICg1MTIgYml0cykNCkV0aGVybmV0IElJLCBTcmM6IEp1bmlwZXJO
  XzM0OjY3OmYwICg0MDphNjo3NzozNDo2NzpmMCksIERzdDogZGE6ODE6NDI6M2Q6ZDA6ZTcgKGRh OjgxOjQyOjNkOmQwOmU3KQ0KSW50ZXJuZXQgUHJvdG9jb2wgVmVyc2lvbiA0LCBTcmM6IDE3OC44
  Mi4xNjkuMCwgRHN0OiAxNTkuODkuMTA0LjE5Mw0KVXNlciBEYXRhZ3JhbSBQcm90b2NvbCwgU3Jj IFBvcnQ6IDUwMDYsIERzdCBQb3J0OiAxMzE4Mg0KUmVhbC1UaW1lIFRyYW5zcG9ydCBQcm90b2Nv bA0KSC4yNjQNCg0KTm8uICAgICBUaW1lICAgICAgICAgICAgICAgICAgICAgICAgICBTb3VyY2Ug ICAgICAgICAgICAgICAgRGVzdGluYXRpb24gICAgICAgICAgIFByb3RvY29sIExlbmd0aCBJbmZv DQogICAgMTQyIDIwMTgtMDQtMTEgMDY6NDA6MDMuMzA2NjgyICAgIDE1OS44OS5YWC5YWCAgICAg ICAgMTkyLjE2OC5YWC5YWCAgICAgICAgIEgyNjQgICAgIDY0ICAgICBQVD1IMjY0LCBTU1JDPTB4
  NUVFOTdDNTUsIFNlcT0zMDU3MiwgVGltZT0zMTkxMjE0MDggU1BTDQoNCkZyYW1lIDE0MjogNjQg Ynl0ZXMgb24gd2lyZSAoNTEyIGJpdHMpLCA2NCBieXRlcyBjYXB0dXJlZCAoNTEyIGJpdHMpDQpF
  dGhlcm5ldCBJSSwgU3JjOiBkYTo4MTo0MjozZDpkMDplNyAoZGE6ODE6NDI6M2Q6ZDA6ZTcpLCBE
  c3Q6IElFVEYtVlJSUC1WUklEXzZlICgwMDowMDo1ZTowMDowMTo2ZSkNCkludGVybmV0IFByb3Rv Y29sIFZlcnNpb24gNCwgU3JjOiAxNTkuODkuMTA0LjE5MywgRHN0OiAxOTIuMTY4LjEuMTg1DQpV
  c2VyIERhdGFncmFtIFByb3RvY29sLCBTcmMgUG9ydDogMTAyOTgsIERzdCBQb3J0OiA1MDIyDQpS
  ZWFsLVRpbWUgVHJhbnNwb3J0IFByb3RvY29sDQpILjI2NA0KUEpTSVAuY29uZjoNCls3MDA0XQ0K
  dHlwZSA9IGVuZHBvaW50DQpjb250ZXh0ID0gaW50ZXJuYWwNCnJld3JpdGVfY29udGFjdCA9IHll cw0KZGlyZWN0X21lZGlhID0gbm8NCnJ0cF9zeW1tZXRyaWMgPSB5ZXMNCjtmb3JjZV9ycG9ydCA9
  IHllcw0KZGlzYWxsb3cgPSBhbGwNCmFsbG93ID0gZzcyMiwgYWxhdywgdWxhdywgZ3NtLCBpbGJj LCBoMjY0DQphb3JzID0gNzAwNA0KYXV0aCA9IGF1dGg3MDA0DQoNCls3MDA0XQ0KdHlwZSA9IGFv cg0KbWF4X2NvbnRhY3RzID0gMg0KDQpbYXV0aDcwMDRdDQp0eXBlPWF1dGgNCmF1dGhfdHlwZT11
  c2VycGFzcw0KcGFzc3dvcmQ9MTIzNA0KdXNlcm5hbWU9NzAwNA0KZXh0ZW5zaW9ucy5jb25mOg0K
  W2ludGVybmFsXQ0KZXh0ZW4gPT4gXzcwMFgsMSxEaWFsKFBKU0lQLyR7RVhURU59KQ0KDQoNCjIw MTgtMDQtMTAgMTY6NDMgR01UKzAyOjAwIEJlbmphbWluIE1hcnR5IDxiZW5qYW1pbi5tYXJ0eUBn bWFpbC5jb208bWFpbHRvOmJlbmphbWluLm1hcnR5QGdtYWlsLmNvbT4+Og0KSSBqdXN0IG5vdGlj ZWQsIHRoZSBjYWxsaW5nIGRldmljZSBpc24ndCBldmVuIHNlbmRpbmcgdGhlIGVhcmx5IG1lZGlh IHZpZGVvIHN0cmVhbS4gSXQganVzdCBzZW5kcyBhbiBlYXJseSBtZWRpYSBhdWRpbyBzdHJlYW0u IElzIHRoZXJlIHByb3BhYmx5IGEgY2hhbmdlIGluIHRoZSBzaWduYWxpbmcgbmVlZGVkPw0KKE9u IGFub3RoZXIgUDJQIFNJUCBTZXJ2ZXIgdGhlIGVhcmx5IG1lZGlhIHZpZGVvIHdvcmtzLikNCg0K
  MjAxOC0wNC0xMCAxMjoyOSBHTVQrMDI6MDAgQmVuamFtaW4gTWFydHkgPGJlbmphbWluLm1hcnR5
  QGdtYWlsLmNvbTxtYWlsdG86YmVuamFtaW4ubWFydHlAZ21haWwuY29tPj46DQpIaSBGbG9yaWFu DQpJIGFscmVhZHkgaGF2ZSB0aGUgZXh0ZXJuYWxfbWVkaWFfYWRkcmVzcyBzZXQgaW4gdGhlIFBK
  U0lQIHNldHVwLiBBbHNvIHRoZSBleHRlcm5hbF9zaWduYWxpbmdfYWRkcmVzcyBpcyBzZXQgdG8g dGhlIFB1YmxpYyBJUC4gSWYgSSBtYWtlIGEgY2FsbCBmcm9tIGFuIEVhcmx5IE1lZGlhICh2aWRl byZhdWRpbykgY2FwYWJsZSBkZXZpY2UgdG8gYW4gRWFybHkgTWVkaWEgY2FwYWJsZSBkZXZpY2Ug KGFsc28gdmlkZW8mYXVkaW8pIHRoZSBFYXJseSBNZWRpYSBhdWRpbyB3b3JrcyBwZXJmZWN0bHku IEJ1dCBubyB2aWRlby4gSWYgSSBzbmlmZiB3aXRoIHdpcmVzaGFyayBvbiB0aGUgcmVjaXBlbnQg ZGV2aWNlIEkganVzdCBzZWUgRzcxMSAoYXVkaW8pIFJUUCB0cmFmZmljLiBUaGUgaDI2NCBSVFAg dHJhZmZpYyBpcyBtaXNzaW5nIGJlZm9yZSBJIGFjY2VwdCB0aGUgY2FsbC4gQWZ0ZXIgYWNjZXB0
  aW5nIHRoZSBjYWxsIHRoZSBoMjY0IFJUUCB0cmFmZmljIGNvbWVzIHRocm91Z2guDQpUaGUgMTgz IFNJUCBwcm90b2NvbGwgY29tZXMgdGhyb3VnaC4gRXZlbiBBc3RlcmlzayBpcyBub3RpY2luZyBp dDoNCi0tIFBKU0lQLzYwMDItMDAwMDAwMTMgaXMgbWFraW5nIHByb2dyZXNzIHBhc3NpbmcgaXQg dG8gUEpTSVAvNjAwMS0wMDAwMDAxMg0KDQpJIHRyaWVkIGJvdGggQXN0ZXJpc2sgMTUgd2l0aCBw anNpcC5jb25mIGNvbmZpZ3VyYXRpb24gYW5kIEFzdGVyaXNrIDEzIHdpdGggc2lwLmNvbmYgKGNo YW5fc2lwKS4gSW4gYm90aCBjYXNlcyBJIGp1c3QgcHV0IHRoZSBib3RoIGNhc2UgQVNUX0ZSQU1F
  X1ZJREVPOiBzdGF0ZW1lbnRzIGJlZm9yZSB0aGUgdHdvIHZvaWNlIGNhc2VzLCBsaWtlIGluIHlv dXIgZGlmZiBhbmQgcmVjb21waWxlZC9yZWluc3RhbGxlZC4NClJlZ2FyZHMNCkJlbmphbWluDQoN
  Cg0KMjAxOC0wNC0xMCA5OjM3IEdNVCswMjowMCBGbG9pbWFpciBGbG9yaWFuIDxmLmZsb2ltYWly QGNvbW1lbmQuY29tPG1haWx0bzpmLmZsb2ltYWlyQGNvbW1lbmQuY29tPj46DQpIaSBCZW5qYW1p biENCg0KWW91J3JlIG9idmlvdXNseSB1c2luZyBhIHNpbWlsYXIgc2NlbmFyaW8gdGhhdCBJIGhh dmUgaW4gcGxhY2UgZm9yIHRlc3RpbmcuDQpJIGluaXRpYWxseSBoYWQgaXNzdWVzIHdpdGggZWFy bHkgbWVkaWEgKG5vdCBvbmx5IHZpZGVvIGFsc28gYXVkaW8pIGFzIHdlbGwgaW4gdGhhdCBzY2Vu YXJpby4gV2hhdCBJIGhhZCB0byBkbyB3YXMgdG8gYWRkaXRpb25hbGx5IHNldA0KDQpleHRlcm5h bF9tZWRpYV9hZGRyZXNzPTx5b3VyIGV4dGVybmFsIElQPg0KDQppbiBwanNpcC5jb25mDQoNCkFs c28sIGFzIEkgd3JvdGUgdGhlIHBhdGNoIGZvciBlYXJseS1tZWRpYSB2aWRlbyBJJ2QgYmUgaW50
  ZXJlc3RlZCBpbiBhbnkgZmVlZGJhY2sgZnJvbSBpdC4NCg0KDQoNCg0KV2l0aCBiZXN0IHJlZ2Fy ZHMNCg0KRmxvcmlhbiBGbG9pbWFpcg0KSW5ub3ZhdGlvbiAtIFNvZnR3YXJlLURldmVsb3BtZW50
  IC0gIFZvSVAgJiBEZXZPcHMNCg0KQ09NTUVORCBJTlRFUk5BVElPTkFMIEdNQkgNCkEtNTAyMCBT
  YWx6YnVyZywgU2FhbGFjaHN0cmHDn2UgNTENClRlbDogKzQzLTY2Mi04NSA2MiAyNQ0KRmF4OiAr NDMtNjYyLTg1IDYyIDI2DQpodHRwOi8vd3d3LmNvbW1lbmQuY29tPGh0dHBzOi8vbGlua3Byb3Rl Y3QuY3VkYXN2Yy5jb20vdXJsP2E9aHR0cCUzYSUyZiUyZnd3dy5jb21tZW5kLmNvbSZjPUUsMSwz LVFGUzc5YmwwN1hKMUF0OS1GTjA0MllXZ19wSWhPb2FNSjNCMTNJekVWc2RVUF8tU0ZaRFVnNXdC
  cm5rRXpRZ0I3VHJaUlF6YWlPMGljU0ozVVhTSlNSbmpJVk91MDY2MUxhLUZqNV9xMUJjelFsUFdV
  X290TSwmdHlwbz0xPg0KDQpTZWN1cml0eSBhbmQgQ29tbXVuaWNhdGlvbiBieSBDb21tZW5kDQoN
  CkZOIDE3ODYxOHogfCBMRyBTYWx6YnVyZw0KDQotLS0tLVVyc3Byw7xuZ2xpY2hlIE5hY2hyaWNo dC0tLS0tDQpWb246IGFzdGVyaXNrLXVzZXJzLWJvdW5jZXNAbGlzdHMuZGlnaXVtLmNvbTxtYWls dG86YXN0ZXJpc2stdXNlcnMtYm91bmNlc0BsaXN0cy5kaWdpdW0uY29tPiBbbWFpbHRvOmFzdGVy aXNrLXVzZXJzLWJvdW5jZXNAbGlzdHMuZGlnaXVtLmNvbTxtYWlsdG86YXN0ZXJpc2stdXNlcnMt Ym91bmNlc0BsaXN0cy5kaWdpdW0uY29tPl0gSW0gQXVmdHJhZyB2b24gSm9zaHVhIENvbHANCkdl c2VuZGV0OiBNb250YWcsIDkuIEFwcmlsIDIwMTggMTg6MTUNCkFuOiBhc3Rlcmlzay11c2Vyc0Bs aXN0cy5kaWdpdW0uY29tPG1haWx0bzphc3Rlcmlzay11c2Vyc0BsaXN0cy5kaWdpdW0uY29tPg0K
  QmV0cmVmZjogUmU6IFthc3Rlcmlzay11c2Vyc10gQXN0ZXJpc2sgYmVoaW5kIE5BVCBFYXJseSBN
  ZWRpYSBWaWRlbw0KDQpPbiBNb24sIEFwciA5LCAyMDE4LCBhdCAxOjA1IFBNLCBCZW5qYW1pbiBN
  YXJ0eSB3cm90ZToNCj4gd29ob28sIHNvIGlmIEkgdW50ZXJzdGFuZCBpdCBjb3JyZWN0bHkgd2l0
  aCB0aGF0IHBhdGNoIGVhcmx5IG1lZGlhDQo+IHZpZGVvIHdvcmtzIG92ZXIgdGhlIEFzdGVyaXNr IHNlcnZlcj8gSW4gb3RoZXIgd29yZHMgdGhlIEFzdGVyaXNrDQo+IHNlcnZlciBnZXQncyBhYmxl IHRvIChwcm9jZXNzLylmb3J3YXJkIHRoZSBlYXJseSBtZWRpYSB2aWRlbyBzdHJlYW0gd2l0aCB0
  aGF0IHBhdGNoPw0KDQpUaGUgcGF0Y2ggZm9yd2FyZHMgdmlkZW8gd2hpbGUgaW4gYW4gZWFybHkg bWVkaWEgc3RhdGUgYmVmb3JlIHRoZSBjYWxsIGlzIGFuc3dlcmVkIGFuZCBicmlkZ2VkLCB5ZXMu DQoNCi0tDQpKb3NodWEgQ29scA0KRGlnaXVtLCBJbmMuIHwgU2VuaW9yIFNvZnR3YXJlIERldmVs b3Blcg0KNDQ1IEphbiBEYXZpcyBEcml2ZSBOVyAtIEh1bnRzdmlsbGUsIEFMIDM1ODA2IC0gVVMg Q2hlY2sgdXMgb3V0IGF0OiBodHRwczovL2xpbmtwcm90ZWN0LmN1ZGFzdmMuY29tL3VybD9hPWh0
  dHBzJTNhJTJmJTJmd3d3LmRpZ2l1bS5jb20mYz1FLDEsZllobzJ0M09HRVBTQzZJTGhWOUlBaGZ5
  cXl2NTdxLWMyZW9kbW1vVGxoUllDbkVwYmdlcXBxWWJrMzloLW1fbERXZmY3VUlsdGQwemFrdjNY
  R2I4NTh5c1ZKYlgwcWVXR3dkc2JjZ3ZkdU5uYUJxVkNEaywmdHlwbz0xICYgd3d3LmFzdGVyaXNr Lm9yZzxodHRwOi8vd3d3LmFzdGVyaXNrLm9yZz4NCg0KLS0NCl9fX19fX19fX19fX19fX19fX19f X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fXw0KLS0gQmFu ZHdpZHRoIGFuZCBDb2xvY2F0aW9uIFByb3ZpZGVkIGJ5IGh0dHBzOi8vbGlua3Byb3RlY3QuY3Vk YXN2Yy5jb20vdXJsP2E9aHR0cCUzYSUyZiUyZnd3dy5hcGktZGlnaXRhbC5jb20mYz1FLDEsWFRv ZW1MZ1B5Nk5RVnliX2RGMXEwcVhTay0zWXlsRjZybUlyV1F2UGhzcHhhZ25GNUc2M1ZIQ1UybkI2
  N1lIalpld01RVTFyVUNNRTRKQlFNRlBtTk9DcGM2RVNPaW5fM0FsNmt0aS1sUm8sJnR5cG89MSAt LQ0KDQpDaGVjayBvdXQgdGhlIG5ldyBBc3RlcmlzayBjb21tdW5pdHkgZm9ydW0gYXQ6IGh0dHBz Oi8vY29tbXVuaXR5LmFzdGVyaXNrLm9yZy8NCg0KTmV3IHRvIEFzdGVyaXNrPyBTdGFydCBoZXJl Og0KICAgICAgaHR0cHM6Ly93aWtpLmFzdGVyaXNrLm9yZy93aWtpL2Rpc3BsYXkvQVNUL0dldHRp bmcrU3RhcnRlZA0KDQphc3Rlcmlzay11c2VycyBtYWlsaW5nIGxpc3QNClRvIFVOU1VCU0NSSUJF
  IG9yIHVwZGF0ZSBvcHRpb25zIHZpc2l0Og0KICAgaHR0cHM6Ly9saW5rcHJvdGVjdC5jdWRhc3Zj LmNvbS91cmw/YT1odHRwJTNhJTJmJTJmbGlzdHMuZGlnaXVtLmNvbSUyZm1haWxtYW4lMmZsaXN0
  aW5mbyUyZmFzdGVyaXNrLXVzZXJzJmM9RSwxLDZWZkpILXlzWXVXcmVsOUFwbDRFcUhiNF9NcERU
  UUhkUTNsSlUzX1pvamdibjRzdFVkTWZjaGxzd1lTU3dWTzlqbW9sLTlINjU4ajJiWnI5Sm1MbWI5
  V0NNNU9YS1RzYl9Ec0JJWUtBQ3RCb3JXUlNVNi1xMUZqSmtyYmMmdHlwbz0xDQoNCi0tDQpfX19f X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19f X19fX19fX18NCi0tIEJhbmR3aWR0aCBhbmQgQ29sb2NhdGlvbiBQcm92aWRlZCBieSBodHRwOi8v d3d3LmFwaS1kaWdpdGFsLmNvbTxodHRwczovL2xpbmtwcm90ZWN0LmN1ZGFzdmMuY29tL3VybD9h PWh0dHAlM2ElMmYlMmZ3d3cuYXBpLWRpZ2l0YWwuY29tJmM9RSwxLFBuTndkRUV5UFk3VTFxYU5j eGI4eVVySlhjcDgtTnZrckRMQ1IzNjR0V25ndHVkbVRMTExLTXFJWk1HU3RUSFpqMTNzbXFmVmg5
  aTVOeGxTS0RyTmN3aFZwVnRYQjJQRXlfcjd2d19tRlFWWHZPVEc2SWoxZ1laVyZ0eXBvPTE+IC0t DQoNCkNoZWNrIG91dCB0aGUgbmV3IEFzdGVyaXNrIGNvbW11bml0eSBmb3J1bSBhdDogaHR0cHM6
  Ly9jb21tdW5pdHkuYXN0ZXJpc2sub3JnLw0KDQpOZXcgdG8gQXN0ZXJpc2s/IFN0YXJ0IGhlcmU6
  DQogICAgICBodHRwczovL3dpa2kuYXN0ZXJpc2sub3JnL3dpa2kvZGlzcGxheS9BU1QvR2V0dGlu ZytTdGFydGVkDQoNCmFzdGVyaXNrLXVzZXJzIG1haWxpbmcgbGlzdA0KVG8gVU5TVUJTQ1JJQkUg b3IgdXBkYXRlIG9wdGlvbnMgdmlzaXQ6DQogICBodHRwOi8vbGlzdHMuZGlnaXVtLmNvbS9tYWls bWFuL2xpc3RpbmZvL2FzdGVyaXNrLXVzZXJzPGh0dHBzOi8vbGlua3Byb3RlY3QuY3VkYXN2Yy5j b20vdXJsP2E9aHR0cCUzYSUyZiUyZmxpc3RzLmRpZ2l1bS5jb20lMmZtYWlsbWFuJTJmbGlzdGlu Zm8lMmZhc3Rlcmlzay11c2VycyZjPUUsMSx1UUNGMGxiLUZWZzdwWGdnS0FVVEV0bkdLMV8xbUd3
  MG5PeFhRdUlqZ3VLbWxuNEZWVXhKeDVVX3pLaVVnakxKZ2E0eVlrVjJnQmtxcG1ZT2F4MTBRelpT
  Y3pTaGZmS01wcUUyaFBoYV9vY0toZHUwVmRibThRLCwmdHlwbz0xPg0KDQoNCg0K

• I added the bind_rtp_to_media_address=yes on all endpoints but still the same behaviour. The funny thing is that the G711 audio early media works and doesn’t have that Private IP issue. I was also able to cross check with chan_sip on Asterisk 15, exactly the same wrong behaviour. See following capture (PJSIP):

  No. Time Source Destination Protocol Length Info
  187 2018-04-11 07:19:56.735967 159.89.XX.XX
  192.168.1.185 H264 943 PT=H264, SSRC=0x3A7AF929, Seq=27144, Time=1248011648 FU-A

  Frame 187: 943 bytes on wire (7544 bits), 943 bytes captured (7544 bits)
  Ethernet II, Src: da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7), Dst:
  IETF-VRRP-VRID_6e (00:00:5e:00:01:6e)
  Internet Protocol Version 4, Src: 159.89.XX.XX, Dst: 192.168.1.185
  User Datagram Protocol, Src Port: 11502, Dst Port: 5022
  Real-Time Transport Protocol H.264

  No. Time Source Destination Protocol Length Info
  188 2018-04-11 07:19:56.735993 159.89.XX.XX
  192.168.1.185 H264 943 PT=H264, SSRC=0x3A7AF929, Seq=27145, Time=1248011648, Mark FU-A End

  Frame 188: 943 bytes on wire (7544 bits), 943 bytes captured (7544 bits)
  Ethernet II, Src: da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7), Dst:
  IETF-VRRP-VRID_6e (00:00:5e:00:01:6e)
  Internet Protocol Version 4, Src: 159.89.XX.XX, Dst: 192.168.1.185
  User Datagram Protocol, Src Port: 11502, Dst Port: 5022
  Real-Time Transport Protocol H.264

  No. Time Source Destination Protocol Length Info
  189 2018-04-11 07:19:56.738966 178.82.XX.XX
  159.89.XX.XX RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x2A1A1C31, Seq=1820, Time=1104983225

  Frame 189: 214 bytes on wire (1712 bits), 214 bytes captured (1712 bits)
  Ethernet II, Src: JuniperN_34:67:f0 (40:a6:77:34:67:f0), Dst:
  da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7)
  Internet Protocol Version 4, Src: 178.82.XX.XX, Dst: 159.89.XX.XX
  User Datagram Protocol, Src Port: 5020, Dst Port: 16130
  Real-Time Transport Protocol

  No. Time Source Destination Protocol Length Info
  190 2018-04-11 07:19:56.738975 178.82.XX.XX
  159.89.XX.XX RTP 214 PT=ITU-T G.722, SSRC=0x49CD55FD, Seq=26679, Time=470333826

  Frame 190: 214 bytes on wire (1712 bits), 214 bytes captured (1712 bits)
  Ethernet II, Src: JuniperN_4f:3f:f0 (40:a6:77:4f:3f:f0), Dst:
  da:81:42:3d:d0:e7 (da:81:42:3d:d0:e7)
  Internet Protocol Version 4, Src: 178.82.XX.XX, Dst: 159.89.XX.XX
  User Datagram Protocol, Src Port: 5004, Dst Port: 18280
  Real-Time Transport Protocol

  2018-04-11 9:11 GMT+02:00 Floimair Florian :

• As I stated previously in order for media to go to the source IP address and port, media has to be received from the endpoint. If this doesn’t happen then you’ll see exactly this behavior – we’ll send to the IP address and port they told us. There’s nothing that Asterisk itself can do in that instance, the endpoint has to send media or place the correct IP address and port in the messages.

  Was any media received from it?

• The current behaviour is that Earlymedia video isn’t working when NAT’s in between are involved. The source/destination IP’s are correct. So the client is sending Early media video + Early media audio to the Asterisk Server “in the cloud” and the Asterisk Server “in the cloud” is sending both to the IP where the Client is located. But strangely just the Early media audio is passing the NAT to the recipent.

  My guess is that the NAT traversal for Early media audio is fine, but the one for Early media video not yet. Can you propably comprehend something in that direction? Or can you guide me to the code part where Asterisk is doing the Port change when a NAT is detected and the Client itself is sending “fake” RTP Early media traffic to get a NAT Binding for incoming RTP Early media traffic?

  Benjamin

  2018-04-11 11:50 GMT+02:00 Joshua Colp :

• The code is in res_rtp_asterisk[1]. It’s not complex and despite the comment is not specific to video. Without logs showing where things are coming from and going I don’t really have anything else to add.

  [1] https://github.com/asterisk/asterisk/blob/master/res/res_rtp_asterisk.c#L6140