Hi!
We’ve run into really interesting issue while testing Kamailio with sipp (15 CPS, 300 calls - 1 failed call, more CPS - more failed calls) with topos involved.
And at some point Kamailio starts to send 2 packets on received INVITE. 1 - with topos applied, 1 - with not.
Main problem is to collect logs at this point, cause load is really high.
Is it possible some internal race condition while topos involved?
PCAP is attached.
I’ve made a picture to illustrate a bit an issue
Regards, Igor
Hi, When you load topos module, at really you added some extra works to kamailio. So the time process is longer than the module is not loaded. What is your the problem in this regards? what do you mean in some race condition? With Regards.Mojtaba
On Thu, 1 Nov 2018 00:10 Igor Olhovskiy, igorolhovskiy@gmail.com wrote:
Hi!
We’ve run into really interesting issue while testing Kamailio with sipp (15 CPS, 300 calls - 1 failed call, more CPS - more failed calls) with topos involved.
And at some point Kamailio starts to send 2 packets on received INVITE. 1
- with topos applied, 1 - with not.
Main problem is to collect logs at this point, cause load is really high.
Is it possible some internal race condition while topos involved?
PCAP is attached.
I’ve made a picture to illustrate a bit an issue
Regards, Igor _______________________________________________ Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
Ok. My answer is here. Seems to be it’s more Homer issue, than Kamailio. sip_trace collect packet before sending. Maybe it’s time to switch to network level of collecting, rather than Kamailio.
Regards, Igor On Oct 31, 2018, 8:23 PM +0200, Igor Olhovskiy igorolhovskiy@gmail.com, wrote:
Hi!
We’ve run into really interesting issue while testing Kamailio with sipp (15 CPS, 300 calls - 1 failed call, more CPS - more failed calls) with topos involved.
And at some point Kamailio starts to send 2 packets on received INVITE. 1 - with topos applied, 1 - with not.
Main problem is to collect logs at this point, cause load is really high.
Is it possible some internal race condition while topos involved?
PCAP is attached.
I’ve made a picture to illustrate a bit an issue
Regards, Igor
-
Igor Olhovskiy -
Mojtaba