olecoot Posted January 8, 2008 Report Share Posted January 8, 2008 We have customers complaining about drop out (packet loss perception in the RTP) during calls. Is this expected with the media CPU being around 25% using version 2.1.4.2345 Win32? Has anyone else experienced this? We have been above 25% Media CPU on older versions without this issue occurring. Our server is Dual CPU 2GB RAM. Will the exe "span" the CPUs or will it use one? Would using the affinity option be of any benefit? Thanks in advance! Quote Link to comment Share on other sites More sharing options...
Vodia PBX Posted January 8, 2008 Report Share Posted January 8, 2008 Affinity: the default is that the process uses CPU 1. I guess you are not using a virtual machine, where CPU affinity becomes "relative"... It would be interesting if the call is in pass-through mode. Then all packets should be forwarded "immediately", this should actually reduc ethe jitter compared to previous versions. Quote Link to comment Share on other sites More sharing options...
olecoot Posted January 8, 2008 Author Report Share Posted January 8, 2008 Affinity: the default is that the process uses CPU 1. I guess you are not using a virtual machine, where CPU affinity becomes "relative"... It would be interesting if the call is in pass-through mode. Then all packets should be forwarded "immediately", this should actually reduc ethe jitter compared to previous versions. No. Not using Virtual Machine. Can you explain what you mean by pass-through mode, please? Quote Link to comment Share on other sites More sharing options...
olecoot Posted January 9, 2008 Author Report Share Posted January 9, 2008 No. Not using Virtual Machine. Can you explain what you mean by pass-through mode, please? Would you suggest disabling pass-through mode? Quote Link to comment Share on other sites More sharing options...
Vodia PBX Posted January 9, 2008 Report Share Posted January 9, 2008 Would you suggest disabling pass-through mode? Maybe worth a try: set allow_pass_through to false. (e.g. http://192.168.1.2/reg_settings.htm?save=s..._through=false). Quote Link to comment Share on other sites More sharing options...
cfcs Posted January 10, 2008 Report Share Posted January 10, 2008 Maybe worth a try: set allow_pass_through to false. (e.g. http://192.168.1.2/reg_settings.htm?save=s..._through=false). You instructed me to folllow these directoin to disable pass-through mode. Try disabling the pass-through mode. This seems to be an ongoing pain in the neck, especially with the CS410 PSTN gateway. You can do this from the web interface (once that you are logged in as admin) with http://1.2.3.4/reg_status.htm?save=save&am...s_through=false. Which way is correct? Quote Link to comment Share on other sites More sharing options...
Vodia PBX Posted January 10, 2008 Report Share Posted January 10, 2008 It does not matter what page you are loading (reg_status or reg_settings). The important part is what you put behind the ?. It is like filling out a form. Because the web interface has no form for pass through mode we need to use this little bit inconvenient way of saving the setting. Quote Link to comment Share on other sites More sharing options...
andrewgroup Posted January 14, 2008 Report Share Posted January 14, 2008 We have been above 25% Media CPU on older versions without this issue occurring. Our server is Dual CPU 2GB RAM. Are you running WIN? if so, I've posted in best practices some of the technical specifications and performance measurements of a reasonably busy mid sized 45 extension system. We've never seen 10% utilization on a single AMD 3100 series processor with a full PRI (23 channels in use). The "Don'ts" are many in regards to maximizing realtime performance for such things as PBX systems and video editing systems. Would it be possible one of the "don'ts" have been violated? (Software Raid control vs. real Disk controller) or something like that. Quote Link to comment Share on other sites More sharing options...
olecoot Posted January 15, 2008 Author Report Share Posted January 15, 2008 Are you running WIN? if so, I've posted in best practices some of the technical specifications and performance measurements of a reasonably busy mid sized 45 extension system. We've never seen 10% utilization on a single AMD 3100 series processor with a full PRI (23 channels in use). The "Don'ts" are many in regards to maximizing realtime performance for such things as PBX systems and video editing systems. Would it be possible one of the "don'ts" have been violated? (Software Raid control vs. real Disk controller) or something like that. Thank you for pointing me in the direction of best practices. Looking at the checklist I would have to say I am in compliance Quote Link to comment Share on other sites More sharing options...
andrewgroup Posted January 15, 2008 Report Share Posted January 15, 2008 Thank you for pointing me in the direction of best practices. Looking at the checklist I would have to say I am in compliance So what do you attribute high CPU utilization to? Is it high with no traffic or does it only rise with traffic? Ulaw vs Alaw, and are you familiar with the default admin codec settings? do they match the ITSP or Gateways? Diagnosing this problem needs to be split 50/50 is this a CPU utilization issue or is this an ITSP QOS issue or are you using POTS or PRI gateways? Are you using two NICS internal and Gateways or Internal and ITSP. Anything you can do to split the problem gets you 50% closer to a solution. (do you have a test ITSP that you can build a test trunk and determine the results?) Are you familiar with Wireshark? other tool? The more you can eliminate with each test the closer you get. That's all for now. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.