webrtc2sip
webrtc2sip copied to clipboard
DoubangoTelecom
webrtc2sip gets into stuck state under load.
What steps will reproduce the problem?
1. Sent 4 simultaneous calls into webrtc2sip Server which automatically stay in
call and then hang up after 30 seconds call duration
2.
3.
What is the expected output? What do you see instead?
Expect that all calls will continue processing. Instead I see that the
webrtc2sip application is still running, TCP sockets get connected, but new
calls are no longer processed.
What version of the product are you using? On what operating system?
VERSION: 2.5.0 on Ubuntu 12.04LTS
Please provide any additional information below.
Individual calls are handled properly. Server only seems to fail under load
with multiple calls.
Original issue reported on code.google.com by [email protected] on 28 May 2013 at 8:14
Attachments:
Noticed several of the following errors
***ERROR: function: "tsip_transport_layer_find()"
file: "src/transports/tsip_transport_layer.c"
line: "849"
MSG: Failed to match response expected to be forwarded via WebSocket transport
***ERROR: function: "tsip_transport_layer_send()"
file: "src/transports/tsip_transport_layer.c"
line: "1009"
MSG: Failed to find valid transport
***ERROR: function: "tdav_video_jb_put()"
file: "src/video/jb/tdav_video_jb.c"
line: "298"
MSG: Too many frames dropped and fps=25
Original comment by [email protected] on 28 May 2013 at 8:18
Hello,
I am having the same issue for 5+ concurrent calls.
***ERROR: function: "tsip_transport_layer_find()"
file: "src/transports/tsip_transport_layer.c"
line: "849"
MSG: Failed to match response expected to be forwarded via WebSocket transport
***ERROR: function: "tsip_transport_layer_send()"
file: "src/transports/tsip_transport_layer.c"
line: "1009"
MSG: Failed to find valid transport
***ERROR: function: "tsip_transport_layer_find()"
file: "src/transports/tsip_transport_layer.c"
line: "849"
MSG: Failed to match response expected to be forwarded via WebSocket transport
***ERROR: function: "tsip_transport_layer_send()"
file: "src/transports/tsip_transport_layer.c"
line: "1009"
MSG: Failed to find valid transport
Sometimes I also come across different debug info (under load):
***ERROR: function: "tnet_get_ip_n_port()"
file: "src/tnet_utils.c"
line: "1130"
MSG: TNET_GET_SOCKADDR has failed with status code: -1
***ERROR: function: "tnet_get_ip_n_port()"
file: "src/tnet_utils.c"
line: "1130"
MSG: (SYSTEM)NETWORK ERROR ==>Socket operation on non-socket
***ERROR: function: "tsip_transport_send_raw_ws()"
file: "src/transports/tsip_transport.c"
line: "374"
MSG: Failed to find peer with local fd equal to 168
***ERROR: function: "tnet_get_ip_n_port()"
file: "src/tnet_utils.c"
line: "1130"
MSG: TNET_GET_SOCKADDR has failed with status code: -1
***ERROR: function: "tnet_get_ip_n_port()"
file: "src/tnet_utils.c"
line: "1130"
Original comment by [email protected] on 18 Jun 2013 at 10:15
@ciprian.dosoftei
5 or 500?
Original comment by [email protected] on 18 Jun 2013 at 10:29
- Changed state: Accepted
That's actually 5, I cannot figure out a pattern yet (quad core 16GB ram
server, plenty of file handles available; webrtc2sip/doubango manually built by
the book).
When the above mentioned transport error occurs it seems webrtc2sip gives up on
it but the client (built on SIPml5) is not aware of it and keeps waiting. I
added a timer which kicks in after a certain amount of time that will
eventually retry the last operation if it hung.
Please let me know if I can provide more information.
Thank you
Original comment by [email protected] on 20 Jun 2013 at 3:37
After getting rid of the overloading issue (by spawning a dedicated webrtc2sip
gateway for each user, works well, no overall server overloading issues) I had
the chance to get to the bottom of the other issues (networking related ones).
I noticed on that particular network there are drops affecting regular plain
websockets in such way that neither the server nor the client are aware of
them. I think it is safe to assume that putting 2+2 together is correct in this
case and the networking issues are causing those warnings (and sometimes call
drops).
Original comment by [email protected] on 21 Jun 2013 at 5:44
We haven't checked this issue yet but using a dedicated server for each user is
not a solution at all.
The error message reported doesn't look to be tied to excessive number of users
Original comment by [email protected] on 21 Jun 2013 at 5:50
I completely agree, it's not elegant at all. When I designed the system and
tested in my office it went smooth. When deployed it was a nightmare and it's
all about those drops. Unfortunately until they fix/replace the LAN, this is my
only choice (I cannot risk a segfault caused by those drops as everyone will
get cut off).
Original comment by [email protected] on 21 Jun 2013 at 5:54
I have the same problem here. One year later! Webrtc2Sip simply stops running
and all the WebRTC users drops.
I would really apreciate some help.
Original comment by [email protected] on 30 May 2014 at 1:56