Hi Carlos
We had similar issues with Asterisk 1.8.11.0 and Asterisk 1.8.32.2 with
timerfd and pthread timing.
The server instance would run for 20 or 30 minutes and then producde legions
of error lines about timerfd or pthread, whichever we were using.
The symptoms were voice quality issues - jitter and packet drops spiked
alarmingly during periods where these errors about timterFD and pthread were
being emitted.
This was also on Centos 7.
The solution in our case was two fold.
- We switched from Asterisk 1.8.11.0 to Asterisk 1.8.32.2
This was step one - since when we used timerfd or pthreads with 1.8.32.3
(even though it was much newer than 1.8.11.0) we STILL had the problem.
- Once on Asterisk 1.8.32.2, we also switched to DAHDI timing
This was step two - when we did this as well, the problems with voice
quality and the tens of thousands of error messages disappeared from the
console and it has been running fine for months since, done about 800 000+
calls so far.
The interesting thing is we have no DAHDI hardware in that system, or any of
the other systems (we run 17 Asterisk instances on Centos 7 real-metal
machines countrywide) where we use DAHDI timing. AFAIK the DAHDI timer is
then emulated in software in the res_timing_dahdi.so module.
So it is not that it is "better" but the upgarding to 1.8.32.2 and
using
DAHDI timing definitely solved our voice quality problems along with making
the constantly repeated timing error messages we were seeing in the console
go away.
Kind regards,
On 4/5/16 3:17 PM, Joshua Colp wrote:> Carlos Chavez wrote:
>> I am currently having a voice quality problem with one of our
>> Asterisk servers. We have checked the network and we have found no
>> problems that could cause the voice to sound cracked and with small
>> interruptions. I am looking at the timing source for Asterisk and it
>> is currently using timerfd even though we have an E1 card installed.
>> Is timerfd better than dahdi? Any recommendations to test if timing
>> may be a problem for voice quality and DTMF?
>
> What is the scenario and the channels involved? Timing is only used
> for things such as playback, music on hold, and ConfBridge. If it's
> strictly a two party call then Asterisk forwards media as received.
>
I am starting to think that the problem may be in the server
hardware itself. We are using a Dell R220 with 8gb of ram and 2 hard disks
in a Raid 1 configuration (Linux Raid). We are using CentOS 7. We had to
remove the raid card from the server to install an E1 card (the raid card
was Windows only so no loss there really). Internally everything sounds
good (from E1 to a conference or music) but once you hit a network interface
we start getting pops and drops.
Anyone with this server and Asterisk ever had issues like these?