You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@httpd.apache.org by Systems Administrator <sy...@gfm.co.uk> on 2002/09/04 10:44:42 UTC
[users@httpd] Apache eats memory and disk space
Does anyone have any insight into this?
We are running 5 production servers all running Apache
and ModSSL. Four of them are also configured with ModPerl
as well.
One of the servers running Apache+ModPerl+ModSSL+OpenSSL
has a few times just eaten the whole of available memory,
1280 Meg, and all swap space >1500MB, and fills up the
error logs with the message "Callback called exit".
I have found that the message "Callback..." was coming from
the ModPerl subsystem, and the relevant documentation says
that this could be the result of very deep recursion, or
possibly a never-ending loop. This would make sense,
however, the sites running on this particular machine,
also run quite happily on the other three servers.
One of which is a 4 processor Xeon III IBM Netfinity
with 1.5GB RAM using Linux kernel 2.4.18-3SMP, another
is a 2 processor Sun Sparc Ultra 420 with 4GB or RAM
running Solaris 8, and finally the other machine is
the same as the one at fault but with 2GB of RAM..
It is a Sun Sparc Ultra-2 dual processor, running
Solaris 8 + recommended patches.
Does anyone know a good way of debugging what's going on, without
putting the machine back into service (it's not reliable
for the client's needs at the moment, although the machine does
run for some time before encountering this situation)
I have looked into rebuilding the server with debug options
set, and using Perl's -DPERL_EMERGENCY_SBRK flag to try
retrieving a stack trace on the trap, however rebuilding
the whole of the Apache system is not an option at the moment.
As a side, does anyone know if you can get away with rebuilding
Mod_Perl only -- without having to rebuild any other part of Apache?
Thanks in advance to anyone that can offer any insight on this,
Cheers,
Con.
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Apache eats memory and disk space
Posted by Hans Juergen von Lengerke <le...@sixt.de>.
It could be that the machine in question has different versions of Perl
modules installed. Nevertheless, you ought to take this question to the
mod_perl list http://perl.apache.org/maillist/modperl.html
You can rebuild mod_perl aside from everything else if you built your
server with 'so' support. Try the -l switch on httpd, it shows the
compiled-in modules. If it shows mod_so.c and doesn't show mod_perl.c
you can rebuild mod_perl without touching the server. Check out the
section on apxs in the mod_perl installation guide.
Regards, Hans
Systems Administrator <sy...@gfm.co.uk> on Sep 4, 2002:
> Does anyone have any insight into this?
>
> We are running 5 production servers all running Apache
> and ModSSL. Four of them are also configured with ModPerl
> as well.
>
> One of the servers running Apache+ModPerl+ModSSL+OpenSSL
> has a few times just eaten the whole of available memory,
> 1280 Meg, and all swap space >1500MB, and fills up the
> error logs with the message "Callback called exit".
>
> I have found that the message "Callback..." was coming from
> the ModPerl subsystem, and the relevant documentation says
> that this could be the result of very deep recursion, or
> possibly a never-ending loop. This would make sense,
> however, the sites running on this particular machine,
> also run quite happily on the other three servers.
> One of which is a 4 processor Xeon III IBM Netfinity
> with 1.5GB RAM using Linux kernel 2.4.18-3SMP, another
> is a 2 processor Sun Sparc Ultra 420 with 4GB or RAM
> running Solaris 8, and finally the other machine is
> the same as the one at fault but with 2GB of RAM..
>
> It is a Sun Sparc Ultra-2 dual processor, running
> Solaris 8 + recommended patches.
>
> Does anyone know a good way of debugging what's going on, without
> putting the machine back into service (it's not reliable
> for the client's needs at the moment, although the machine does
> run for some time before encountering this situation)
>
> I have looked into rebuilding the server with debug options
> set, and using Perl's -DPERL_EMERGENCY_SBRK flag to try
> retrieving a stack trace on the trap, however rebuilding
> the whole of the Apache system is not an option at the moment.
>
> As a side, does anyone know if you can get away with rebuilding
> Mod_Perl only -- without having to rebuild any other part of Apache?
>
> Thanks in advance to anyone that can offer any insight on this,
>
>
> Cheers,
>
> Con.
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP Server Project.
> See <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org