apachectl restart broken for prefork

[Jeff Trawick <tr...@bellsouth.net>]

(maybe this should read "still broken"?)

Why are we still using signals sent to the children in the
non-graceful restart case?  That doesn't interrupt select or mutex
acquire, as described previously.

ap_mpm_pod_signal() seems to be busted anyway.  We loop writing to the
pod (why loop?) until it fails due to pipe full, then we bail out
without waking up the child via connect().

Working on it...  just wondering if there is a reason why non-graceful
shouldn't call ap_mpm_pod_signal() too...

-- 
Jeff Trawick | trawickj@bellsouth.net | PGP public key at web site:
       http://www.geocities.com/SiliconValley/Park/9289/
             Born in Roswell... married an alien...

Re: apachectl restart broken for prefork

[rb...@covalent.net]

On Thu, 7 Jun 2001, Jeff Trawick wrote:

> (maybe this should read "still broken"?)
>
> Why are we still using signals sent to the children in the
> non-graceful restart case?  That doesn't interrupt select or mutex
> acquire, as described previously.
>
> ap_mpm_pod_signal() seems to be busted anyway.  We loop writing to the
> pod (why loop?) until it fails due to pipe full, then we bail out
> without waking up the child via connect().
>
> Working on it...  just wondering if there is a reason why non-graceful
> shouldn't call ap_mpm_pod_signal() too...

Using the pod can't do non-graceful.  The point of non-graceful, is that
it is immediate.  If all of your threads are currently serving requests,
then sending data to the pod won't do anything until at least one request
is done.

Ryan

_______________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------