I had installed Opensips in my laptop. I am using Sipp as user agent for UaC and UaS and i had connect to swicth.
192.168.1.99:5060 192.168.1.100:5060 192.168.1.101:5060
I am testing dialog transaction, I had a problem that my Opensips is not forwarding ACK and BYE messages to UaS .So on UaS, i am getting failure calls, But on UaC and i am getting successful calls. I dont know what is going inside and i had observed on wireshark that opensips is not sending ACK and BYE messages to UaS.Please help me on this problem and i am very much thankfull to all.
your sipp xml must handle record-route/route header.
thanks for the reply
I had question, How many maximum number of calls can be handle by opensips.Because when i had tested it by 100 cps and maximum limit for calls is 10,000 only 9997 and i got 2 calls failed and keep it increasing by 200 ,300 , 400 ,500,600,700 cps for 10,000 calls and i got failure rate by 9998,9982,7576,7020,6262,4859.I had observed that for 100 cps the failure rate is 9997 and for 200 cps it was 9998 and remaining calls, the failure rate is decreased by uniformly 9982,7576,7020,6262,4859.How can judge the maximum number calls can be handled by opensips.While i had observed in wireshark, i had seen 408 response not the sever error and there are unexpected message like 408 more number for more calls per second.whether it is correct or not ,there should not be unexpected message while testing. Please give me any solution regarding this
opensips performance depends on many factors, such as hardware, os, os configurations, opensips configurations, and test call scenarios. in our test, the results are uniform. 100cps 100% completed, 200 cps, 98% completed. there are some metrics can be used to evalue the results, 1. cpu/memory usage. 2. network usage. 3. opensips memory usage. 4. opensips number of threads. in your case, if the cpu/memory/network usages did not reach limits, then it should be opensips configuration issue.
First of all, I would like to thank you very much for replying and also for your fast response to my question.I am using ubuntu 10.04 and i had observed in the system monitor for the above two cases you said 1. cpu/memory usage 2.network usage for 100 to 600 cps its keep varying and for 600 it reached 100 percent and Network usage around 2 to 3 Mbps.
While coming to memory usage i had kept 128 Mb in the path /etc/default/opensips.Whether it is correct or not i dont know please comment on this.
One important thing i need to mention i am getting 408 error for some failure calls for testing 100cps to 600 Cps.Whether it is correct or not .My question i need to get busy error or 500 serer error to judge for server failure .because i am testing server performance using basic scenario like invite -->180-->200->Ack->bye.Please give me some idea regarding this and i am very much thank full to you.
i mentioned to monitor the usages means to see if the usages and performance are linear. If they are not, there must be some issues. at least when usages < 90%, they should be linear.
for opensips, you may try "-m 512" option, set "children=64", and disable any feature you do not use, such as dns.
for 408, it looks like sipp uas end issue. it is interesting because i expect uac end reaches limits first. did sipp report any issue such as max number of open files? you may use multiple sipp uac and uas.
How are you ,i had observed it and it is linear and i had made some change to the opensips file like you told that children=64
but i am getting the same error . I didn't understand on thing in your previous msg that you had mention one parameter that is "-m 512" and i am little bit confused whether your saying about the memory value which need to be change from 128 to 512 mb.Becuase i am using sipp, So in sipp we will fix it for maximum number of calls.whether your speaking about the thing which i mention or else your talking about a new thing and which i am unable to understand it.please give me some and your suggestion are helping me a lot.
i am confused. you said it is linear. then where is the issue? for example, 100cps 100% completed->50% cpu/memory usage, 200cps 100% completed->60% cpu/memory usage, …, 600cps 100% completed->100% cpu/memory usage. "-m 512" is to use 512M shared memory.