DarkKnight Posted January 7, 2010 Report Posted January 7, 2010 We found a bug in the current version 3.4.0.3201 (Linux) that a significant amount of Extension-to-Extension calls on the same domain will not give ringback to the originating Extension (all the originating party hears is dead air, so the originating party has no idea what has happened to the call). In every case the destination Extension successfully rings and can pickup the Extension-to-Extension call, and the call then proceeds as normal. We have made Wireshark captures of normal calls with ringback to the originating Extension and calls where the ringback fails to be made to the originating extension. The captures show to be exactly the same. We tested the current version 3.4.0.3201 (Linux) and previous versions for this issue and found the following Extension-to-Extension ringback failure rates: Version 3.1.0.3031 (Linux) ---5% Version 3.2.0.3143 (Linux) ---3% Version 3.3.1.3177 (Linux) ---37% Version 3.4.0.3201 (Linux) ---24% This needs to work RELIABLY, this is something that has not been a problem in previous versions, yet somehow when "new" versions come out, functions that have always worked in previous versions are broken. Quote
Vodia PBX Posted January 7, 2010 Report Posted January 7, 2010 It could be a race condition with the loading from the file system. Especially when you have a network-mounted file system, it could be that the loading is too slow and that somehow leads to a stall in the playback procedure. If you have a lab setup, send a email to support with the indication of your OS. It is definivively worth trying out the 3.5 build. Quote
DarkKnight Posted January 8, 2010 Author Report Posted January 8, 2010 It could be a race condition with the loading from the file system. Especially when you have a network-mounted file system, it could be that the loading is too slow and that somehow leads to a stall in the playback procedure. If you have a lab setup, send a email to support with the indication of your OS. It is definivively worth trying out the 3.5 build. -----The OS is Red Hat EL5. We do not run NFS, so if you mean a remotely mounted disk array--- no that is not our environment. If you mean a hosted environment, then yes that is what we have. Can you get me that 3.5 build, please? Quote
hosted Posted January 10, 2010 Report Posted January 10, 2010 We found a bug in the current version 3.4.0.3201 (Linux) that a significant amount of Extension-to-Extension calls on the same domain will not give ringback to the originating What phones are you using? Quote
olecoot Posted January 11, 2010 Report Posted January 11, 2010 We seem to be having issues similar to this along with intermittent AA playback as posted in another thread. http://forum.pbxnsip.com/index.php?showtopic=3222&hl= We would like to try the 3.5 version for RedHat EL5 please. Quote
DarkKnight Posted January 11, 2010 Author Report Posted January 11, 2010 What phones are you using? All manners of Polycom phone models, using the latest firmware. This is NOT a Polycom issue: we have all manners of Polycom phones also running on pbxNsip Version 3.1.0.3031 (Linux) & Version 3.2.0.3143 (Linux) with *little-to-no instances* of this problem. Listing the Polycom models used would be wasting everyone's time and taking this issue in the exact wrong direction, and wasting more time, instead of focusing on what in these newer builds was broken. Quote
Vodia PBX Posted January 13, 2010 Report Posted January 13, 2010 All manners of Polycom phone models, using the latest firmware. This is NOT a Polycom issue: we have all manners of Polycom phones also running on pbxNsip Version 3.1.0.3031 (Linux) & Version 3.2.0.3143 (Linux) with *little-to-no instances* of this problem. Listing the Polycom models used would be wasting everyone's time and taking this issue in the exact wrong direction, and wasting more time, instead of focusing on what in these newer builds was broken. Well, yes and no... The way the subscription is handled was changed; and it might have not been such an obvious problem before. The problem was that the phone subscribe for a list of dialog events, and the PBX now answers with "sorry no, but try again in one hour". The phones try again after three seconds (maxbe a unit problem, s vs. ms). We had other installations where this problem was fixed in version 3.5. In this version we answer, "okay" but send a empty list. That slows down the traffic significantly. BTW it should be very easy to see in the traces (or Wireshark) if there is a subscription storm. We should look at a Wireshark tracve anyway to see what is going on there. Quote
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.