asterisk users - Dec 2005 - FOLLOWUP: .call files on PRI / Zombie AGI proces ses in FC2 / 1.2 Beta 1 under load

Following up on 2 recent posts of mine:

".call files on PRI not waiting for answer in desired context"

-and-

"Zombie AGI processes in FC2 / 1.2 Beta 1 under load"

both issues experienced under 1.2 beta 1

I can confirm that both issues went away when I downgraded back to 1.0.9.
Don't know why, I'm just documenting it for the list in case someone
else
experiences these issues. I will try the same thing again over the holidays
under 1.2 -release and see if I can duplicate it. Text of original posts:


".call files on PRI not waiting for answer in desired context"

If I generate a .call file to an external callee through my PRI, Asterisk
will not wait to execute the priority in the target context, and instead
will continue on as soon as the channel is dialled. I want it to wait for an
answer, THEN continue. It detects the answer correctly. I have
callprogress=yes in Zapata.conf. I have read the wiki with respect to this
issue. 

Weird thing is, I swear this worked the way I wanted it to when I was
running 1.0.9, now it's not under 1.2 beta 1. Am I crazy?


"Zombie AGI processes in FC2 / 1.2 Beta 1 under load"

I have an extremely simple AGI script like this:

#!/bin/bash
ISODATE=`date -iso-8601=seconds`
echo "SET VARIABLE ISODATE \"$ISODATE\""

The script's intent is to return the current date and time back to Asterisk
in an ISO format as a variable. It works fine. Calling the script from
Asterisk returns the variable correctly. 

Under load (> a dozen or so concurrent calls) the script still works but it
tends to zombie. Sometimes it does, sometimes it doesn't. I've seen some
background:

http://lists.digium.com/pipermail/asterisk-users/2003-August/017310.html

http://lists.digium.com/pipermail/asterisk-users/2003-August/017202.html

But this is FC2, latest SMP kernel / 1.2 Beta 1 . The zombies build up and
appear not to affect call quality but I'd like to eliminate them. Anyone
have any tips? tia