Hi Liviu, You will definitely want a very low logging level (log_level = 1) I have set log level as 3, I will try by changing to 1 and check.
You should print the output of "opensips -V", maybe improvements can be made Am using OpenSIPS version 2.2.2. Choosing the right compile flags can also help Am using the default compile flags. Should anything be changed in that in order to improve the performance? To conclude, please try to understand the bottleneck first (by using "top", I have used top command to see the bottleneck, the CPU load average is 4.5 max, out of 8CPU's and memory usage is 2.8% out of 16GB. I will keep continue testing, will let you know if I come across any bottlenecks. Regards, Agalya From: [email protected] [mailto:[email protected]] On Behalf Of Liviu Chircu Sent: Monday, October 24, 2016 11:09 AM To: [email protected] Subject: Re: [OpenSIPS-Users] OpenSIPS performance. Hi, Agalya! A successful stress test should highlight a bottleneck that we are no longer able to easily fix. Common bottlenecks with any network server application will always include: - CPU - I/O (disk / network) Most often, disk writes are to blame, either through stderr or syslog. You will definitely want a very low logging level (log_level = 1) when doing thousands of CPS, as even a single log line per request can lower the overall throughput. Choosing the right compile flags can also help. You should print the output of "opensips -V", maybe improvements can be made. To conclude, please try to understand the bottleneck first (by using "top", "netstat", "vmstat", etc.), then we can start working on it! Liviu Chircu OpenSIPS Developer http://www.opensips-solutions.com On 24.10.2016 17:38, Ramachandran, Agalya (Contractor) wrote: Hi Bogdan, Am using opensips 2.2.2 version and using opensips as only proxy. In the default script, in route[relay], I have called setdsturi(); No other changes with the default script. Earlier, I have used only the default value of S_MEMORY and P_MEMORY in the /etc/default/opensips. With this default values, for the end to end established call with pause of (24 to 45 sec), I could reach only 150cps. When I tried to increase S_MEMORY and P_MEMORY to 1024 and 64 respectively, and having opensips children =12, I could achieve 800 cps. I have also re-tuned OpenSIPS log file, by adding "-/var/log/opensips.log". When I see opensips performance with default script I see you are achieving 9000cps. What are the places I need to take a look to tune up still, to get more performance. Your expertise would help me a lot. Am running on VM, with 8 core CPU, 16GB RAM and processor is Intel Xeon E312xx(Sandy Bridge). Regards, Agalya _______________________________________________ Users mailing list [email protected]<mailto:[email protected]> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
_______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
