Hi Prashanth, as you were talking about load test, I was suspecting too many tcp sessions to be open between just two IP addresses and a single port at server side per unit of time, which can cause problems when any particular tcp port at client side is reused for a new tcp session earlier than the post-session guard timer elapses. The number of calls which are active in parallel at a given moment is not important in this case. What matters is the total number of tcp sessions established during last 60 (or maybe 120?) seconds as compared to the total number of tcp ports which SIPp is using at client side.
But this does not explain why another pair of virtual machines has no issue, as I suppose you've installed them from the same image (i.e. with the same configuration) like the problematic ones. From what you wrote now it is not clear whether after the change to tcp mono mode the issue still exists on the original pair of virtual machines. In order to localize the cause of the issue, I'd recommend to capture the ethernet traffic on both machines to a file (using tcpdump or dumpcap) and then analyse the captures (especially tcp sessions' establishment in tcp multisocket mode) using Wireshark. This should show both how many tcp ports are used at client side and whether there is an issue with tcp session establishment, packet loss or something else. Pavel Dne 25.11.2015 v 15:59 Prashanth Devarajappa napsal(a): > Hi Pavel, > > There are 2 systems(VMs) involved, client and server. Client is where I have > SIPp running pushing calls to server @ 5060. > Yes, I believe my scenario is using single IP for both local and remote > address. I have not seen more than 45 calls at a given time for my scenario. > > I believe this issue to be somehow related to server receiving the calls. Ie > SIPp's socket write fails due to remote end. How can I prove this ? > Until now, all these error occur during night. > > For time being I have changed below items > > * Switched to mono socket(-t t1) from multi socket(-t tn) > * enabled 'trace_calldebug' > > Additional Info: > I have second set of test environment(with 2 different VMs for client and > server) where same scenario works fine w/o any issue. > > Regards > Prashanth > > > Date: Tue, 24 Nov 2015 13:52:06 +0100 > From: sindelka <sinde...@ttc.cz> > Subject: Re: [Sipp-users] SIPp errors during load test > To: sipp-users@lists.sourceforge.net > Message-ID: <56545d76.30...@ttc.cz> > Content-Type: text/plain; charset="windows-1252" > > Hello Prashanth, > > how many individual IP addresses are involved in the test? Is your SIPp UAC > scenario using a single local IP address to send all the INVITEs to a single > remote IP address? If so, how many sessions (calls) are in progress when the > error occurs? > > Pavel > > ------------------------------------------------------------------------------ Go from Idea to Many App Stores Faster with Intel(R) XDK Give your users amazing mobile app experiences with Intel(R) XDK. Use one codebase in this all-in-one HTML5 development environment. Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs. http://pubads.g.doubleclick.net/gampad/clk?id=254741551&iu=/4140 _______________________________________________ Sipp-users mailing list Sipp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sipp-users
