8.2.3 - maybe a problem
Mark.Andrews at nominum.com
Mark.Andrews at nominum.com
Wed Jul 5 02:44:29 UTC 2000
> Date: Tue, 04 Jul 2000 16:57:22 +1000
> From: Mark.Andrews at nominum.com
> Message-ID: <200007040657.QAA19701 at drugs.dv.isc.org>
>
> | The following is awaiting inclusion to 8.2.3-REL / 8.2.3-TB6.
>
> Thanks. That explains the kill(-1, SIGTERM) - it doesn't explain
> how all the zombie proceses are being left around (which for me)
> seems to be the trigger that causes the vfork() to fail in the first
> place (ie: swap running out). Or if it does, I don't see how.
All I can think of is that SIGCHLD is not getting delivered and
hence the appropriate need is not being set. The race condition
that used to exist no longer does (despite the comment) as the
only thing the signal handler does is set the bit.
The waitpid() is executed on the main stack and we loop looking
for all dead children so even if we loose a signal subsequent
signals should allow things to be cleaned up.
It's almost like SIGCHLD is getting permanently blocked.
Mark
>
> But thanks fo rthe fix, it should certainly cure the most serious
> problem...
>
> kre
--
Mark Andrews, Nominum Inc.
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: Mark.Andrews at nominum.com
More information about the bind-workers
mailing list