Re: Apache = CPU hog? (fwd)

[Randy Terbush <ra...@zyzzyva.com>]

We've had to deal with this exact issue on Solaris, although it
has not been a problem in Apache (yet).


> Rob Hartill wrote:
> > 
> > 
> > Anyone have any ideas?
> 
> I wonder... I read this slightly worrying stuff in man signal on SCO 5...
> 
>           The signal catching function is not executed during certain slow
>           processes even though a signal has been caught. Slow processes are
>           considered to be read(S), write(S), open(S), or ioctl(S) system
>           calls on a slow device (like a terminal; but not a file), a pause
>           (S) routine, or during a wait(S) routine that does not return
>           immediately due to the existence of a previously stopped or zombie
>           process. Then the interrupted system call may return a -1 to the
>           calling process with errno set to EINTR.
> 
> sigaction() does not have this qualification. Could this be the problem?
> 
> I'm fairly sure I've seen SCO 5 fail to clear up the children, too.
> 
> Cheers,
> 
> Ben.
> 
> > 
> > > Further to my posting, here is an example of a session (for apache 1.1b2 
> > > on Solaris 2.5) I think this shows adequately what happens - and I waited
> > > quite a while between commands.  Would be a pain for more than a few 
> > > servers, and could mean the thing is down for a few minutes while I
> > > do this.  We actually have a test server which we "swap" with the live
> > > one when we want to change over.
> > > 
> > > 
> > > Cheers
> > > Keep up the fab work
> > > Mike