[Linux-ha-dev] [PATCH]The problem that child process of Apache does not stop

[HIDEO YAMAUCHI]

Hi,

> Hi,
> 
> On Tue, Dec 11, 2007 at 05:34:00PM +0900, HIDEO YAMAUCHI wrote:
> > Hi,
> > 
> > > Hi,
> > > 
> > > On Mon, Dec 10, 2007 at 03:53:56PM +0900, HIDEO YAMAUCHI wrote:
> > > > Hi,
> > > > 
> > > > There are the following problems to a resource script of current apache.
> > > > (http://developerbugs.linux-foundation.org//show_bug.cgi?id=1800)
> > > > 
> > > > 1.A parent process of apache falls.
> > > 
> > > You mean fails ...
> > 
> > Yes.
> > 
> > > 
> > > > 2.In the stop script, child process is not stopped.
> > > 
> > > ... and the children are left hanging around.
> > 
> > Yes.
> > 
> > > 
> > > > 3.Though the process of the parent falls, the child process
> > > > replies and continues handling it.
> > > 
> > > I see your point, it's just strange that nobody ever complained
> > > about this. Under which circumstances does this happen?
> > 
> > When only an owner process failed, it is generated.
> > can confirm it when downed by a Kill command artificially.
> >   
> >   ex. kill -9 ApachePID  
> 
> I see. But does that ever happen in normal operation?
> 
> Also, isn't it that the children never listen for requests, but
> they may be handling existing requests or waiting for the parent
> process to pass one?
> 
> In the former case, perhaps it's a good idea to let them finish:
> given the nature of http they will be short-lived and the user
> experience is going to be better. I think that these should not
> be killed and they are going to leave anyway, probably within a
> subsecond period.
> 
> As for the latter, those are the preforked processes for faster
> start. I'm not sure what happens to them, but the important
> question is if they would prevent the new apache instance from
> starting. I guess not. At the worst case, they may just remain
> and be swapped out until some merciful soul removes them.
> 
> Can somebody who's experience apache administrator help here?
> 
> Thanks,
> 
> Dejan

Probably I think that do not get up by the normal operation.

But, it is a problem to fail in the reboot of the resource in the same node by this child process.
The failure of the reboot of this resource continues till child process stops.

It is the problem that may be hard to happen, but thinks that I am effective.

Regards,
Hideo Yamauchi.

> 
> 
> 
> > > 
> > > > I made a patch to stop all processes.
> > > > Evaluate this patch, and please include it.
> > > 
> > > Thanks for the patch!
> > > 
> > > Dejan
> > > 
> > > > 
> > > > Regards,
> > > > Hideo Yamauchi.
> > > 
> > > 
> > > > _______________________________________________________
> > > > Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> > > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> > > > Home Page: http://linux-ha.org/
> > > 
> > > _______________________________________________________
> > > Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> > > Home Page: http://linux-ha.org/
> > > 
> > 
> > _______________________________________________________
> > Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> > Home Page: http://linux-ha.org/
> _______________________________________________________
> Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> Home Page: http://linux-ha.org/
>