Ubuntu
php5 package

Mailing from php5 module fails when number of Apache VirtualHosts is above +/-1500

Bug #419191 reported by Hark
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
php5 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: libapache2-mod-php5

My shared webserver suddenly stopped mailing yesterday. Both the PHP mail() function or calling /usr/sbin/sendmail directly from exec() or something similar gives this error in /var/log/exim4/mainlog (of course /usr/sbin/sendmail is just a symlink to exim4):

2009-08-26 13:35:51 1MgGn9-0002rA-Tm == <email address hidden> R=smarthost T=remote_smtp_smarthost defer (-1): smtp transport process returned non-zero status 0x000b: terminated by signal 11

After a long search I finally found that mailing starts working again when I delete a few random Apache VirtualHosts. I can really not explain why this happens, but I can reproduce this behaviour every time. The VirtualHosts configuration is certainly correct, and Apache nor PHP gives any error message. The mail is added to the exim4 queue correctly as far as I know, but it is not sent and frozen immediately. With exim -M message_id on the command line I am able to send each message afterwards without a problem. I tried different exim4 configurations (without smarthost for example), but that doesn't give any change.
On another server I use suPHP (CGI), with exactly the same versions of Apache and Exim4, there this problem does not exist. Also on the server where I have this problem a Perl CGI script that calls /usr/sbin/sendmail runs just fine, and sends the mail. So I think it must be some bug in libapache2-mod-php.

Description: Ubuntu 8.04.3 LTS
Release: 8.04

libapache2-mod-php5:
  Installed: 5.2.4-2ubuntu5.7
  Candidate: 5.2.4-2ubuntu5.7
  Version table:
 *** 5.2.4-2ubuntu5.7 0
        500 http://nl.archive.ubuntu.com hardy-updates/main Packages
        500 http://security.ubuntu.com hardy-security/main Packages
        100 /var/lib/dpkg/status
     5.2.4-2ubuntu5 0
        500 http://nl.archive.ubuntu.com hardy/main Packages

Revision history for this message
Ondřej Surý (ondrej) wrote :

Have checked your memory using memtest? Signal 11 could be a sign that there is something wrong with the hardware. Especially if the same configuration on other host is ok.

Another option is to replicate whole server to some virtual (kvm) host and try if it happens there as well.

I am afraid that without more information (gdb backtrace), and step-by-step guide how to reproduce the error, nobody will be able to help you.

Revision history for this message
Hark (ubuntu-komkommerkom) wrote : Re: [Bug 419191] Re: Mailing from php5 module fails when number of Apache VirtualHosts is above +/-1500

To be sure I just started the kvm on another host. I experience exactly
the same problem there. This is not a hardware fault.

Mark Hunting wrote:
> Well in fact this server is already inside a kvm, so I don't think
> hardware is the problem. The configuration on the other host is OK
> because I don't use PHP as a module there. The other host is also
> running inside a kvm on exactly the same hardware.
> A gdb backtrace is a bit too complicated for me I'm affraid. But to
> reproduce the error it should be enough to just create more than 1500
> different VirtualHosts, and then use the mail() function in PHP5. If
> more info is needed please ask.
>
> Ondřej Surý wrote:
>
>> Have checked your memory using memtest? Signal 11 could be a sign that
>> there is something wrong with the hardware. Especially if the same
>> configuration on other host is ok.
>>
>> Another option is to replicate whole server to some virtual (kvm) host
>> and try if it happens there as well.
>>
>> I am afraid that without more information (gdb backtrace), and step-by-
>> step guide how to reproduce the error, nobody will be able to help you.
>>
>>
>>
>
>

Revision history for this message
Hark (ubuntu-komkommerkom) wrote :

Sorry for mixing up some email addresses. My second e-mail didn't come
through because of that. Here is it again:
Well in fact this server is already inside a kvm, so I don't think
hardware is the problem. The configuration on the other host is OK
because I don't use PHP as a module there. The other host is also
running inside a kvm on exactly the same hardware. A gdb backtrace is a
bit too complicated for me I'm affraid. But to reproduce the error it
should be enough to just create more than 1500 different VirtualHosts,
and then use the mail() function in PHP5. If more info is needed please ask.

Ondřej Surý wrote:
> Have checked your memory using memtest? Signal 11 could be a sign that
> there is something wrong with the hardware. Especially if the same
> configuration on other host is ok.
>
> Another option is to replicate whole server to some virtual (kvm) host
> and try if it happens there as well.
>
> I am afraid that without more information (gdb backtrace), and step-by-
> step guide how to reproduce the error, nobody will be able to help you.
>
>

Revision history for this message
Hark (ubuntu-komkommerkom) wrote :

Sorry for bothering you again. I'm marking this bug as invalid. When I
install another MTA (postfix, sendmail, esmtp, ...) everything works.
It's very strange that Exim4 stops working with lots of Apache
virtualhosts using php5 as module, but it will be way too complicated
for me to find the cause of this, and I don't think its a bug in
libapache2-mod-php5.

Changed in php5 (Ubuntu):
status: New → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.