ID: 15896
Updated by: [EMAIL PROTECTED]
Reported By: [EMAIL PROTECTED]
-Status: Open
+Status: Feedback
Bug Type: PostgreSQL related
Operating System: Redhat Linux
PHP Version: 4.1.2
-Assigned To:
+Assigned To: yohgaki
New Comment:
I don't see this kind of behavior.
If memory is broken, the cause may be anywhere.
Could you build with min. config options?
./configure --with-apxs --without-mysql --with-pgsql --enable-debug
Could you paste shortest possible reproducible script with
the configuration that I mentioned?
Make sure you check apache error log also.
Previous Comments:
------------------------------------------------------------------------
[2002-03-25 06:25:38] [EMAIL PROTECTED]
Hi,
This issue is very reproduceable here also, on two very different
setups sharing only Apache and Postgresql versions.
(Apache 1.3.20, 22 and 23 and Postgresql 7.1.3)
The setups involved that reproduce the reported bug are one production
system setup (running Postgresql and Apache separately, Apache 1.3.20 /
1.3.22 and PHP 4.0.6patched)
and additionally one internal test environment (running Postgresql and
Apache on the same server, Apache 1.3.23 and PHP 4.1.2).
The OS is Linux, but distributions are Debian Potato with 2.2 kernel
(production) and Slackware 8.0 with 2.4 kernel (test env.).
Production system has very high load, test environment has little to no
load. Both reproduce the same issue as described in the bugreport.
Additional info, my PHP is a lot less complex, I configured with :
./configure
--enable-trans-sid
--with-config-file-path=/etc/httpd
--with-apxs=/usr/local/apache/current/bin/apxs
--with-pgsql
--without-mysql
This is a very serious issue to me, my production servers are behaving
less than well, very possibly because of this issue (frequent memory
leaks / apache instability).
Any feedback highly appreciated !
Regards
Denis Braekhus
------------------------------------------------------------------------
[2002-03-06 08:31:34] [EMAIL PROTECTED]
Hi,
I've got a reproduceable problem with PHP when using
PostgreSQL. The problem does not appear when running
MySQL. I suspect a buffer overflow of some kind.
A good example of the problem follows, code taken from my
db library class and the query function:
$myVariable = "My Value";
$limit = -1;
$offset = 0;
print( "MyVar: $myVariable <br>" );
Will normally output:
MyVar: My Value<br>
However; after 55-60 database calls with the function it
will print:
MyVar: 0<br>
The 0 is the value assigned to the $offset variable. If I
remove the $offset = 0; line the output will be:
MyVar: -1<br>
As you can see all variables will be assigned to the same
value ( after n runs ).
It allways happens at the same place in the code even
after apache restarts etc. I have two different scrips
using the db library reproducing the same problem. And
this does not happen when using the mysql implementation
of the db library.
It may be something in the area of; a buffer for storing
the variable assignments overflows due to a buffer
overflow in the postgresql library. But this is only a
guess since I don't know the internals of PHP/PostgreSQL.
PHP configure line:
'./configure' '--with-dom' '--with-xml'
'--with-apxs=/usr/sbin/apxs' '--enable-ftp'
'--enable-trans-sid' '--with-config-file-path=/etc/httpd'
'--with-mysql=shared,/usr' '--with-pgsql=shared,/usr'
'--enable-inline-optimization' '--with-ttf' '--with-gd'
'--enable-gd-native-ttf' '--with-imap' '--includedir=/usr'
'--with-openssl=/usr' '--with-zlib-dir=/usr'
'--with-ldap=shared,/usr' '--with-openssl=shared,/usr'
'--enable-dbg=shared' '--with-dbg-profiler'
'--with-kerberos'
System:
Linux dozer.ez.no 2.4.9-31 #1 Tue Feb 26 07:11:02 EST 2002
i686 unknown
Server API:
Apache
I'm _not_ running APC or Zend accellerator.
There are no error messages in the log and no segfaults.
Regards
B�rd Farstad
------------------------------------------------------------------------
--
Edit this bug report at http://bugs.php.net/?id=15896&edit=1
