Hi Olivier, We've not done stress testing specifically on the websockets interface, so I can't say how many reg/sec Bono can handle.
Looking at the logs below, the REGISTER that's come in has a CSeq of 432550. Later on there is: [processing of another request arriving from sprout] => 04-06-2014 15:05:32.056 Debug pjsip: sip_endpoint.c Processing incoming message: Response msg 200/REGISTER/cseq=432550 (rdata0xcd2e28) This has the same CSeq as the request, so it looks like a response to the original REGISTER, not a different one. Are the Call-IDs different in this case? Can you send me the Bono log? Ellie From: DURECU, OLIVIER (OLIVIER) [mailto:[email protected]] Sent: 09 June 2014 19:14 To: Eleanor Merry Cc: [email protected] Subject: RE: Messages dropped by bono There is not so much to see in network traces. Some messages arrive at Bono and never get out of it... The number of dropped messages increases as the number of register per sec increases (one at 10 msg/sec, 3 or 4 @ 15,....) There are no error messages as you can see in my piece of log. For lost messages, Bono just stops processing them after having read them from the websocket. It's like its attention is caught by another message coming from sprout and it forgets what it was doing with the received register... Using the websocket interface, how many register/sec is Bono able to handle ? Le 9 juin 2014 18:21, Eleanor Merry <[email protected]<mailto:[email protected]>> a écrit : Hi Olivier, Bono can support more than 10 reg/sec. Are the REGISTERs definitely being routed to Sprout? Can you get network trace? Is Sprout rejecting the REGISTERs due to overload? You would see lines like "Rejected request due to overload" in the Sprout logs if so. Ellie -----Original Message----- From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of DURECU, OLIVIER (OLIVIER) Sent: 04 June 2014 16:22 To: [email protected]<mailto:[email protected]> Subject: [Clearwater] Messages dropped by bono Hi I made some tests where 10 REGISTER were done quickly (within 1 sec) on websockets. And systematically 1 or 2 of them fail... The REGISTER arrives at Bono and never follow its ways to sprout... there is no error, it just happens nothing for this message A piece of logs for a concerned message 04-06-2014 15:05:32.006 Debug websockets.cpp:407: Validating incoming web socket connection 04-06-2014 15:05:32.006 Debug websockets.cpp:411: Client requested subprotocol sip - agreeing 04-06-2014 15:05:32.006 Debug websockets.cpp:430: New web socket connection, creating PJSIP transport 04-06-2014 15:05:32.006 Debug websockets.cpp:205: Incoming connection from 159.217.144.101:29478 04-06-2014 15:05:32.006 Verbose pjsip: ws Websockets (null) started, published address is 10.90.0.254:443 04-06-2014 15:05:32.006 Debug websockets.cpp:437: Created WS transport 04-06-2014 15:05:32.032 Debug websockets.cpp:450: Received message from websockets 04-06-2014 15:05:32.032 Debug websockets.cpp:453: Sending message to PJSIP... 04-06-2014 15:05:32.032 Debug pjsip: sip_endpoint.c Processing incoming message: Request msg REGISTER/cseq=432550 (rdata0x7f85180120a0) 04-06-2014 15:05:32.032 Verbose stack.cpp:226: RX 577 bytes Request msg REGISTER/cseq=432550 (rdata0x7f85180120a0) from WS 159.217.144.101:29478: --start msg-- REGISTER sip:[email protected] SIP/2.0^M Call-ID: [email protected]^M<mailto:[email protected]%5eM> CSeq: 432550 REGISTER^M From: "user4" <sip:[email protected]>;tag=0^M To: "user4" <sip:[email protected]>^M Via: SIP/2.0/WS hk7t1amgpvgku870.invalid:5082;branch=z9hG4bK8wo2xFdsyWfrCSgwnsOd3XAuX1TOhtyv,SIP/2.0/WS hk7t1amgpvgku870.invalid:5082;branch=z9hG4bK8wo2xFdsyWfrCSgwnsOd3XAuX1TOhtyv^M Max-Forwards: 70^M Contact: "user4" <sip:[email protected]>^M Expires: 120^M User-Agent: Java client^M Organization: Bell-labs-vllx^M Supported: path^M Content-Length: 0^M ^M --end msg-- 04-06-2014 15:05:32.032 Debug stack.cpp:411: Queuing cloned received message 0x7f8518016578 for worker threads 04-06-2014 15:05:32.032 Debug websockets.cpp:456: Passed message to PJSIP successfully 04-06-2014 15:05:32.032 Debug stack.cpp:189: Worker thread dequeue message 0x7f8518016578 04-06-2014 15:05:32.032 Debug pjsip: sip_endpoint.c Distributing rdata to modules: Request msg REGISTER/cseq=432550 (rdata0x7f8518016578) 04-06-2014 15:05:32.032 Debug pjsip: tsx0xcfcf48 Incoming Request msg REGISTER/cseq=432550 (rdata0x7f8518016578) in state Trying 04-06-2014 15:05:32.032 Debug stack.cpp:191: Worker thread completed processing message 0x7f8518016578 04-06-2014 15:05:32.032 Debug stack.cpp:197: Request latency = 77us [processing of another request arriving from sprout] => 04-06-2014 15:05:32.056 Debug pjsip: sip_endpoint.c Processing incoming message: Response msg 200/REGISTER/cseq=432550 (rdata0xcd2e28) Strange, isn't it? I guess Bono is able to support more than 10 registration / sec in the stress tests you are doing ? Olivier DURECU Bell Labs / IP Platform Research Villarceaux Center Tel: +33(0) 160402759 GSM: +33(0) 683776482 _______________________________________________ Clearwater mailing list [email protected]<mailto:[email protected]> http://lists.projectclearwater.org/listinfo/clearwater _______________________________________________ Clearwater mailing list [email protected] http://lists.projectclearwater.org/listinfo/clearwater
