[Asterisk-Dev] For those exploring the my_zt_write() EAGAIN issue...

[Kris Boutilier]

I've noticed this evening that on my systems at least, the issue seems to occur far more frequently if the zaptel card is slaved to a remote timing source (ie. span=1,1,0). If the card is internally clocked (ie. span=1,0,0) the rate of EAGAIN errors goes way, way down and accordingly the audiable pops and clicks also vanish. There is no noticeable change in interrupt performance between the two configurations according to zttest or zttest-mod and no interrupts report as dropped in /proc/zaptel/1

This is an Asus P4BG-MX, running a Celeron 2.4 and 256Mb of ram with one t100p card and two Ethernet cards. Kernel is 2.6.11.8 with Ingo Molnars 'realtime' patches applied. Priority of the Interrupt servicing thread for the t100p card has been raised to be the highest priority process (-52, using chrt) and asterisk (cvs-head from 3 may) is running with the priority boost flag (-p) applied. As a test load 30 channels were opened, 15 IAX2 (7 to one host, 3 to another) and 15 PRI channels. The system was transcoding each call pair between g726 and ulaw. Load average was averaging roughly 1.94 during the tests. Some additional notes are in pastebin: http://pastebin.ca/11045

Hope this helps someone out there get closer to the bottom of the issue. The difference in audio quality is enough that I'm about to drive around and reverse the timing relationships on all my PBXes. Shame I can't feed my telco too.

Kris Boutilier
Information Services Coordinator
Sunshine Coast Regional District