You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by David Reid <da...@jetnet.co.uk> on 2000/03/22 18:54:30 UTC

Logs in restart

I've narrowed the problem with restarts on BeOS down to the logs.  When 
it tries to open the error_log for the restart it fails and then exits.  
I guess this isn't a problem on other platforms, but does anyone have 
any immeadiate ideas?

d.

Re: Logs in restart

Posted by David Reid <dr...@jetnet.co.uk>.
I found the problem for restarts on BeOS was due to problems with the
child_setup code (file permissions when opening a file but user setup in a
more general sense).  It took a while to get to as the error wasn't being
printed due to the problems and the lack of a error reporting at that point
in the restart :(  Still, I found a few more errors along the way :)

d.
----- Original Message -----
From: "Martin Pool" <mb...@linuxcare.com>
To: <ne...@apache.org>
Cc: <da...@jetnet.co.uk>
Sent: Thursday, March 23, 2000 7:03 PM
Subject: Re: Logs in restart


> On Thu, 23 Mar 2000 rbb@apache.org wrote:
>
> > Probably.  I posted a suggested fix for that problem, but never
> > implemented it.  The basic problem, is that we assume when we duplicate
> > files that the same pool(context) should be used for both files.  This
is
> > a bogus assumption.  We need to use a different pool for the duplicate
> > version that we are creating.
>
> That sounds good to me.  I will look at implementing it.
>
> --
> Martin Pool, Linuxcare, Inc.
> +61 2 6262 8990
> mbp@linuxcare.com, http://www.linuxcare.com/
> Linuxcare. Support for the revolution.
>
>


Re: Logs in restart

Posted by Martin Pool <mb...@linuxcare.com>.
On Thu, 23 Mar 2000 rbb@apache.org wrote:

> Probably.  I posted a suggested fix for that problem, but never
> implemented it.  The basic problem, is that we assume when we duplicate
> files that the same pool(context) should be used for both files.  This is
> a bogus assumption.  We need to use a different pool for the duplicate
> version that we are creating.

That sounds good to me.  I will look at implementing it.

-- 
Martin Pool, Linuxcare, Inc.
+61 2 6262 8990
mbp@linuxcare.com, http://www.linuxcare.com/
Linuxcare. Support for the revolution.


Re: Logs in restart

Posted by rb...@apache.org.
On Thu, 23 Mar 2000, Martin Pool wrote:

> On Wed, 22 Mar 2000, David Reid wrote:
> 
> > I've narrowed the problem with restarts on BeOS down to the logs.  When 
> > it tries to open the error_log for the restart it fails and then exits.  
> > I guess this isn't a problem on other platforms, but does anyone have 
> > any immeadiate ideas?
> 
> A week ago I posted about a problem of logs being closed by a pool as the
> parent exits.  Could this be related?

Probably.  I posted a suggested fix for that problem, but never
implemented it.  The basic problem, is that we assume when we duplicate
files that the same pool(context) should be used for both files.  This is
a bogus assumption.  We need to use a different pool for the duplicate
version that we are creating.

Ryan

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


Re: Logs in restart

Posted by Martin Pool <mb...@rproxy.sourceforge.net>.
On Wed, 22 Mar 2000, David Reid wrote:

> I've narrowed the problem with restarts on BeOS down to the logs.  When 
> it tries to open the error_log for the restart it fails and then exits.  
> I guess this isn't a problem on other platforms, but does anyone have 
> any immeadiate ideas?

A week ago I posted about a problem of logs being closed by a pool as the
parent exits.  Could this be related?

--
Martin


RE: Logs in restart

Posted by "William A. Rowe, Jr." <wr...@lnd.com>.
> From: David Reid [mailto:david@jetnet.co.uk]
> Sent: Wednesday, March 22, 2000 11:55 AM
> 
> I've narrowed the problem with restarts on BeOS down to the 
> logs.  When 
> it tries to open the error_log for the restart it fails and 
> then exits.  
> I guess this isn't a problem on other platforms, but does anyone have 
> any immeadiate ideas?

Wasn't closed in the first process?

A hook chain isn't being followed in the restart process?
We hook into open_logs, but I can't seem to find any hook for
shutdown (not declared via DECLARE_HOOK).  It would seem we need
a buffer flush and close if you are buffering upon restart.

I remember seeing issues with Win32 not completing error reports,
and bet this is related.