You are viewing a plain text version of this content. The canonical link for it is here.
Posted to modperl@perl.apache.org by Boysenberry Payne <bo...@humaniteque.com> on 2005/08/17 21:05:31 UTC
Apache (mod_perl) restart issues
When I restart apache I get the following errors on child processes:
child process 424 did not exit, sending another SIGHUP
One for each child process. After the above I get:
SIGHUP received. Attempting to restart
I'm using mod_perl 1.26 and apache 1.3.33 (only happens with my
mod_perl scripts running)
I thought at first it was DBI connections not getting closed properly,
they seem to be fine though.
I'm not exactly sure how to go about tracking down what's preventing
the original SIGHUP.
Any ideas?
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
Re: Apache (mod_perl) restart issues
Posted by Boysenberry Payne <bo...@humaniteque.com>.
On Aug 19, 2005, at 6:45 PM, Todd Finney wrote:
> perldoc perldiag
Attempt to free unreferenced scalar
(W internal) Perl went to decrement the reference count of a
scalar
to see if it would go to 0, and discovered that it had
already gone
to 0 earlier, and should have been freed, and in fact,
probably was
freed. This could indicate that SvREFCNT_dec() was called
too many
times, or that SvREFCNT_inc() was called too few times, or
that the
SV was mortalized when it shouldn't have been, or that
memory has
been corrupted.
For someone like me ( 4 months of perl experience ) this is rather hard
to make
heads or tails of. Until 4 months ago I was a php guy. I imagine this
is during the
"clean up" phase of mod_perl shutting down since this is when I get the
errors.
So here is what doesn't make sense to me:
How do I trace which references are having it happen?
What does a scalar being mortalized mean?
How can I tell if the memory is being currupted?
And, since this is on a server I can ssh into, but don't have root
access for
how can I go about debugging it the best without causing my host too
much
of a headache?
Thanks for all the help,
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
STDOUT clones in mod_perl
Posted by David Scott <ds...@earthlink.net>.
I've been trying to pass the STDOUT filehandle to subroutines, with
mixed success. Seems I should be able to do:
my $fh = \*STDOUT;
my_sub( $fh );
where my_sub is:
sub my_sub {
my ( $fh ) = @_;
print $fh "foobar\n";
}
I'm not getting any output at all, in fact the Apache process blocks.
Also,
tie $fh, ref $r, $r
complains that TIESCALAR can't be found in class Apache, although this
is supposed to work.
I'm using Apache 1.3.27 and mod_perl 1.
Any ideas?
David
Re: Apache (mod_perl) restart issues
Posted by Todd Finney <tf...@boygenius.com>.
At 06:37 PM 8/19/2005 -0500, Boysenberry Payne wrote:
>On OSX 10.3.9:
>perl5 (revision 5.0 version 8 subversion 1 RC3)
>mod_perl 1.26 / apache 1.3.33
>
>In error_log as:
>[warn] child process 1354 still did not exit, sending a SIGTERM
>for each child process then finally:
>[notice] caught SIGTERM, shutting down
>regardless of whether it's a restart or stop command
[http://modperlbook.org/html/ch05_03.html]
>On External Linux 2.4.21-32.0.1.EL:
>perl5 (revision 5.0 version 8 subversion 0)
>mod_perl 1.26 / apache 1.3.33
>
>In the error_log as:
>null: Attempt to free unreferenced scalar.
perldoc perldiag covers this particular error message.
Re: Apache (mod_perl) restart issues
Posted by Boysenberry Payne <bo...@humaniteque.com>.
On Aug 19, 2005, at 4:10 PM, Philip M. Gollucci wrote:
> I believe what you are seeing is a known bug in the Apache 1.3.x/mp1
> series.
>
> I don't think its your code.
>
> I believe google searches will yeild this; I know I've seen it before.
I've found a lot of people mentioning similar errors when I search via
google.
I haven't found out whether it's something that's a real concern or not.
Nor mention of a specific bug.
How can I find a list of known bugs?
I seem to be having two, probably from the same thing, but
appearing different on each platform.
On OSX 10.3.9:
perl5 (revision 5.0 version 8 subversion 1 RC3)
mod_perl 1.26 / apache 1.3.33
In error_log as:
[warn] child process 1354 still did not exit, sending a SIGTERM
for each child process then finally:
[notice] caught SIGTERM, shutting down
regardless of whether it's a restart or stop command
On External Linux 2.4.21-32.0.1.EL:
perl5 (revision 5.0 version 8 subversion 0)
mod_perl 1.26 / apache 1.3.33
In the error_log as:
null: Attempt to free unreferenced scalar.
It's setup on a cron job to restart at a regular interval during
development.
It gets a whole mess of the errors, 5 child processes worth.
I think they might go away with warnings off right? Either way it makes
me wonder what I should do about it?
Thanks,
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
Re: Apache (mod_perl) restart issues
Posted by "Philip M. Gollucci" <pg...@p6m7g8.com>.
Boysenberry Payne wrote:
> So I figured out how to get the perl debugger working with mod_perl,
> thanks to
> Practical mod_perl. I still don't know how to track down the scalars
>> [Sun Feb 13 01:00:00 2000] null: Attempt to free unreferenced scalar.
>> ...<snip>
>>> I'm using mod_perl 1.26 and apache 1.3.33 (only happens with my
>>> mod_perl scripts running)
I believe what you are seeing is a known bug in the Apache 1.3.x/mp1
series.
I don't think its your code.
I believe google searches will yeild this; I know I've seen it before.
--
END
------------------------------------------------------------
What doesn't kill us can only make us stronger.
Nothing is impossible.
Philip M. Gollucci (pgollucci@p6m7g8.com) 301.254.5198
Consultant / http://p6m7g8.net/Resume/
Senior Developer / Liquidity Services, Inc.
http://www.liquidityservicesinc.com
http://www.liquidation.com
http://www.uksurplus.com
http://www.govliquidation.com
http://www.gowholesale.com
Re: Apache (mod_perl) restart issues
Posted by Perrin Harkins <pe...@elem.com>.
On Fri, 2005-08-19 at 18:40 -0500, Boysenberry Payne wrote:
> I turned off both:
>
> PerlFreshRestart On
Don't ever use that.
> PerlModule Apache::StatINC
> PerlInitHandler Apache::StatINC
Only use that in development, not on a production server.
Apache::Reload is the newer and better version.
- Perrin
Re: Apache (mod_perl) restart issues
Posted by Boysenberry Payne <bo...@humaniteque.com>.
I turned off both:
PerlFreshRestart On
PerlModule Apache::StatINC
PerlInitHandler Apache::StatINC
on the Linux server and still get the errors.
I'm going to look into Devel::Leak and Devel::LeakTrace
Thanks,
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
On Aug 19, 2005, at 4:12 PM, Perrin Harkins wrote:
> On Fri, 2005-08-19 at 16:02 -0500, Boysenberry Payne wrote:
>> I still don't know how to track down the scalars
>> being leaked.
>> Could they be being caused from something like the following?
>>
>> sub example {
>> return 1;
>> }
>> example;
>
> No, I don't think so. You can try using things like Devel::Leak and
> Devel::LeakTrace to find the problem. I've never used them personally.
>
> I always do "apachectl stop && apachectl start" to restart my servers.
> Graceful restarts don't work well on all systems and sometimes cause
> memory leaks.
>
> - Perrin
>
>
>
Re: Apache (mod_perl) restart issues
Posted by Perrin Harkins <pe...@elem.com>.
On Fri, 2005-08-19 at 16:02 -0500, Boysenberry Payne wrote:
> I still don't know how to track down the scalars
> being leaked.
> Could they be being caused from something like the following?
>
> sub example {
> return 1;
> }
> example;
No, I don't think so. You can try using things like Devel::Leak and
Devel::LeakTrace to find the problem. I've never used them personally.
I always do "apachectl stop && apachectl start" to restart my servers.
Graceful restarts don't work well on all systems and sometimes cause
memory leaks.
- Perrin
Re: Apache (mod_perl) restart issues
Posted by Boysenberry Payne <bo...@humaniteque.com>.
So I figured out how to get the perl debugger working with mod_perl,
thanks to
Practical mod_perl. I still don't know how to track down the scalars
being leaked.
Could they be being caused from something like the following?
sub example {
return 1;
}
example;
Because the the void context of the returned value? Maybe I'm
confused...
On the advise of someone on another list I stripped my code down to
just the handler
and I still get restart errors on my OS X box, which is the only server
of the two I have
root access on.
Another oddity, on the OSX box I only get the errors on apachectl
restart, not
stop or graceful.
Any help would be much appreciated
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
On Aug 17, 2005, at 2:46 PM, Boysenberry Payne wrote:
> The last set of errors are happening on my OSX box.
> I get a different set of errors on our Linux server when it restarts:
>
> [Sun Feb 13 01:00:00 2000] null: Attempt to free unreferenced scalar.
> ...<snip>
>
> I get a whole mess of those, then I get:
>
> Scalars leaked: 234
>
> I'm not sure how to go about fixing these error on either server
>
>
> Boysenberry
>
> boysenberrys.com | habitatlife.com | selfgnosis.com
>
> On Aug 17, 2005, at 2:05 PM, Boysenberry Payne wrote:
>
>> When I restart apache I get the following errors on child processes:
>>
>> child process 424 did not exit, sending another SIGHUP
>>
>> One for each child process. After the above I get:
>>
>> SIGHUP received. Attempting to restart
>>
>> I'm using mod_perl 1.26 and apache 1.3.33 (only happens with my
>> mod_perl scripts running)
>>
>> I thought at first it was DBI connections not getting closed
>> properly, they seem to be fine though.
>>
>> I'm not exactly sure how to go about tracking down what's preventing
>> the original SIGHUP.
>>
>> Any ideas?
>>
>> Boysenberry
>>
>> boysenberrys.com | habitatlife.com | selfgnosis.com
>>
>>
>>
>
>
>
Re: Apache (mod_perl) restart issues
Posted by Boysenberry Payne <bo...@humaniteque.com>.
The last set of errors are happening on my OSX box.
I get a different set of errors on our Linux server when it restarts:
[Sun Feb 13 01:00:00 2000] null: Attempt to free unreferenced scalar.
...<snip>
I get a whole mess of those, then I get:
Scalars leaked: 234
I'm not sure how to go about fixing these error on either server
Boysenberry
boysenberrys.com | habitatlife.com | selfgnosis.com
On Aug 17, 2005, at 2:05 PM, Boysenberry Payne wrote:
> When I restart apache I get the following errors on child processes:
>
> child process 424 did not exit, sending another SIGHUP
>
> One for each child process. After the above I get:
>
> SIGHUP received. Attempting to restart
>
> I'm using mod_perl 1.26 and apache 1.3.33 (only happens with my
> mod_perl scripts running)
>
> I thought at first it was DBI connections not getting closed properly,
> they seem to be fine though.
>
> I'm not exactly sure how to go about tracking down what's preventing
> the original SIGHUP.
>
> Any ideas?
>
> Boysenberry
>
> boysenberrys.com | habitatlife.com | selfgnosis.com
>
>
>