On Mon, Jul 15, 2013 at 7:56 PM, mayamatakeshi <[email protected]>wrote:
> > > > On Mon, Jul 15, 2013 at 11:49 AM, mayamatakeshi > <[email protected]>wrote: > >> >> >> >> On Mon, Jul 15, 2013 at 10:13 AM, mayamatakeshi >> <[email protected]>wrote: >> >>> Hello, >>> we are using opensips 1.6 r. 7055 under CentOS 5.3 x86_64 and this works >>> fine on several machines. >>> I am tasked with preparing upgrade to CentOS 6.4 x86_64 so I have >>> prepared a Virtual Machine to run this OS, installed opensips and used the >>> same opensips.cfg . >>> >>> However, I am seeing the following issue: reception of "100 Trying" >>> doesn't cancel module tm fr_timer but if other provisional responses like >>> "180 Ringing" is received, fr_timer is canceled. >>> I am sure the "100 Trying" is received by opensips because it doesn't >>> retransmit the INVITE. >>> >> >> Sorry, by mistake I looked at the wrong capture. Indeed, opensips is >> retransmitting the INVITE. But checking Call-ID, from-tag and Via headers, >> they match and so, opensips should not do retransmission. It seems somehow >> it is ignoring the "100 Trying". >> >> So to further investigate I enabled debug with "opensipsctl fifo debug 4" >> (instead of default debug 0), but surprisingly, now opensips accepts the >> "180 Trying" (doesn't retransmit the INVITE) and fr_timer is canceled. >> > > In the above I meant "100 Trying". > > >> I tested several times and just by switching debug off and on is enough >> to cause the problem to appear/disappear. >> >> >> >>> Everything else seems to be fine. >>> >>> Does anyone has any idea why this could be happening? >>> Also, to confirm I have prepared 3 CentOS 6.4 x86_64 VMs, but in one of >>> them the problem doesn't happen but I don't know why this is so as i >>> compared the packages installed (yum list installed) and they are all the >>> same, including version numbers. >>> >> >> Also, on the machine where the problem doesn't manifest itself, it is the >> opposite: >> if I enable debug with "opensipsctl fifo debug 4", fr_timer fails to be >> canceled by "100 Trying". >> > > Retrying several times, the problem occasionally happens on this machine > too. So it seems to be some race condition. > > I was able to try with latest revisions of 1.7 and 1.8 branches and the > problem also happens with them. > In my tests, I have this scenario: sipp uac --> INVITE --> opensips --> FreeSWITCH sipp uac <-- opensips <-- 100 Trying <-- FreeSWITCH however, if I replace FreeSWITCH with a sipp uas, the problem doesn't happen. I took care of preparing the sipp uas to generate the "100 Trying" exactly the same as FS would do it, but still I could not reproduce the problem. So i think the presence of FS has some strange influence in the environment and i cannot be sure the problem is with opensips itself anymore. Sorry for the noise. Regards, Takeshi
_______________________________________________ Users mailing list [email protected] http://lists.opensips.org/cgi-bin/mailman/listinfo/users
