Stefan Viljoen
2019-Jan-15 06:04 UTC
[asterisk-users] Various extensions ring once and go to voicemail - Thomas Peters
Here’s what I get: apbx*CLI> module show like timing Module Description Use Count res_timing_pthread.so pthread Timing Interface 0 res_timing_dahdi.so DAHDI Timing Interface 4 2 modules loaded So what would you suggest? (And thanks in advance.) Thomas I've had some good experience with res_timing_dahdi both when we ourselves were still on 1.8 and now with us on Asterisk 13 as well. To force usage of a certain timer, specify in your modules.conf, e. g. to force use of DAHDI timing only, I did the following in my modules.conf: . . . load => res_timing_dahdi.so noload => res_timing_pthread.so noload => res_timing_timerfd.so That said, we have had some weird issues trying to run Asterisk in virtual machines - all our instances (16 of them) are physical machines. We did a deployment at Azure in a Centos 7 "stock Azure" VM awhile ago and it suddenly lost the capability to encode .gsm audio files. All .gsm files the virtualised Asterisk 13 instances produced were all corrupt and no player would want to play the .gsm files. Neither could SOX convert them to anything. So we had to switch over to .wav, and then use a mixmonitor hook and manually convert the .wav files back to .gsm in SOX after each recording was written by Asterisk in .wav format. There were no errors logged, Asterisk just mysteriously lost the capacity to encode .gsm files when running on the Azure VM instance we had. So quite probably the virtual environment / hypervisor you're using is part of the issue and switching timing modules around won't solve anything... Have you checked that the system time is sane, and that one second on a stop watch externally to the VM instance, equates to one second inside it? Because the symptoms described could indicate that the clock in the VM is just running too fast - or that some timing implementation detail inside Asterisk itself is running too fast. Regards Stefan