Hello,
Do not contradict the expert.
You must find out the parent process of the zombies. In the output
from "ps axl" commands, this is shown in the PPID column (use "ps
-el" on Solaris and other SYS5-ish platforms).
The parent process is not cleaning up as it should.
Wietse
Oh no, i do not contradict but sometimes a correct answer can be too general
to be helpful the fast way. sometimes you need a second jump-start ;) sorry,
its not easy...
i did now abandon the attempt to use a shellscript as policy-server. its an
interesting way but not really comfortable. I still think my script is ok
but was stopped by some kind of serverlimit for unprivileged users (pipe
limit etc.) But it doesnt really matter any more, i moved now to perl.
In postfix sources is greylist policy-service example included which could
easily customized to fit my needs. there is only one question to postfixs
behaviour.
it seems that postfix spawns the perl-script and is using it multiple
times. I think command_time_limit is starting to count at first startup of
the servicechild.
In some cases i find in log following line:
process id xxxxx: command time limit exceeded
by verbose logging of processstart and errorline i can see that message is
written $command_time_limit seconds (default) after starting the
policychild. I do not really know how to handle this.
my function shows result with nearly no loss of time and script was used a
lot of times until this child brings the warning but i can not influence
how often script is used by postfix.
Can somebody explain this problem please?
Thanks,
Andre
