Hi, I fixed the "nonce" error with the latest 1.4.2 svn release. Previously, I got the latest release mixed up with old .so file. Now it is working fine. Thank you for the fix.
The only problem I have left is about the performance of opensips when Auth is involved. Without Authentication, Opensips can process >10K registration requests/s. When Authentication is involved, it can only handle 80 registration requests/s. The box in this case is a Pentinum dual core with 2G Ram. After I moved the mysql DB to a qual-core Xeon 4G RAM box, cps of registration requests increased to 550/s. Is this roughly that I should be expecting? I hope to see the kind of benchmarking number that other people are seeing. Thank you in advance for your help. Regards, JB On Sat, Dec 6, 2008 at 11:32 AM, Juan Backson <[EMAIL PROTECTED]> wrote: > Hi, > > I am running load testing on opensips using sipp. When tested with pure > registration without Auth ( opensips just store info in the location table), > I get cps of >10K. However, when I change the script to enable auth via > database table, I get only cps of around 80. Does anyone know why? I am > using mysql and I don't think mysql should perform that bad. Can someone > give me some pointers on where to look to identify the bottleneck? > > The other problem I am having is that I am getting with the latest 1.4.2 > svn: > Dec 6 11:09:53 localhost openser-1.2.0-ul[9514]: ERROR:auth:build_auth_hf: > no more nonces can be generated > Dec 6 11:09:53 localhost openser-1.2.0-ul[9512]: ERROR:auth:challenge: > failed to generate nonce > > How can I fix it? > > Another problem I am struggling with is that when testing with sipp with > authentication turned on, sipp responses with: > 2008-12-06 19:15:26:890 1228562126.890990: Aborting call on UDP > retransmission timeout for Call-ID '[EMAIL PROTECTED]'. for some of the > calls even though I also set the recv_timeout to 10s: > ./sipp 192.168.1.105:5030 -sf tests/reg_with_auth/xmlregister.xml -inf > tests/reg_with_auth/register.csv -r 20000 -m 100000 -trace_rtt -trace_screen > -l 10 -trace_err -max_retrans 0 -recv_timeout 10s > > With recv_timeout of 10s, I don't think sipp should even timeout at all. > Does anyone know what I may have done wrong? > > > Thank you in advance for all your kind help. > > Best Regards, > JB > > > > > > > > > > > > > > >
_______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
