ID: 20449
Comment by: dave at flitsservice dot nl
Reported By: josh at zebotech dot com
Status: Open
Bug Type: Session related
Operating System: redhat 7.3
PHP Version: 4.4.0-dev
New Comment:
* PHP Version 4.2.2 (and 4.1.1)
* Apache 2.0.40 (and 1.3.23)
* Windows NT 5.1 build 2600 (Windows XP sp1)
Allright, I experience this problem also! (thnx Google). It drives me
mad, I already did a downgrade to PHP v4.1.1 without succes.
I experience the same situation as jpmarray, the page looks like it
loads, I get some HTML code and a part of my page is displayed...
Suddenly it breaks off the output showing some unfinished HTML code and
a "Page cannot be displayed" occurs.
And I'm talking of the well known phpinfo.php now! The strange thing
which someone else mentioned also here, is that it works perfectly well
with all non-IE related browsers. I tested it for example with Mozilla
1.0 and everything is working fine.
And now comes the strange part. I have another server here, running
exactly the same OS (Windows NT 5.1 build 2600) and exactly the same
installation and php.ini file (version 4.1.1) and that is working
perfectly well. The only difference is the Apache server version;
1.3.23
And I don't want to downgrade Apache unless I have too...
Oh and another thing, I tried to run phpinfo.php on different machines,
all resulting with the same problem. No go with IE and Mozilla and
Opera 7 everything looks OK...
But while I test this when I'm writing this now, I see that it isn't OK
either with other browsers... It does not result in a "page cannot be
found" error but it just unfinishes the page ending. This looks like
it's doing it at random... Sometimes stopping earlier and sometimes
near the end of the output...
Very strange! Nothing is wrong with my php maximum execution time or
something. Remember that I have two different servers with exactly the
same PHP installation...
Maybe I'll try an Apache downgrade to exclude some things out
tomorrow... First I'll get some sleep, luckily this is less critical
than the problem of Josh...
Dave Lolkes de Beer
The Netherlands
Previous Comments:
------------------------------------------------------------------------
[2003-01-31 14:57:30] jpmurray at hxti dot com
Hello all,
* PHP Version 4.2.3
* Apache 1.3.27
* Linux
* config: ./configure' '--with-apxs'
'--with-oci8=/usr/local/oracle/m01/app/oracle/product/8.1.7/'
'--with-openssl' '--with-curl' '--with-mysql'
I'm having this problem every day, and can reproduce at will.
Extremely frustrating.
For me it only occurs upon (successful) login to my site. Instead of
going to the php index, I get a "Page cannot be displayed". I'm using
non-cookie sessions managed by an Oracle database.
Once I am in, past the login, the error never occurs however. It seems
only to be a problem when the session is first initialized via
session_start(). After this, I pass the sid via the URL, and its
subsequently checked by the target script against the oracle db for
authenticity- No problems at all there. Does anybody else see this
problem occur only upon doing things such as logging in?
Thanks,
J. Murray
HX Technologies
------------------------------------------------------------------------
[2003-01-08 11:45:20] phpBug20449 at rehner dot net
Further testing shows that on Windows 2000 Server with Identical
Configurations as the Windows 2000 Pro (I did diffs to be sure)
reveals:
Win2000Server 2 Processor- 2 failues in an estimates 300,000 tests.
Win2000Server 4 processor - 0 failures in 1000 tests (Just started
running this one.)
Basically I had everyone in the office run the test against the 2
Processor system at the same time. Only 2 failures in that many
requests I can live with. So is it more of an issue on less capable
systems?
------------------------------------------------------------------------
[2003-01-08 01:52:58] josh at zebotech dot com
interesting.
However, I switched my session manager to a dbase function and it still
died. How would file locking effect that?
I'm interested in the fact though that you can actually witness the bug
first hand. I only saw that it was happening. However, I could never
get the thing to happen to me.
Also, since I have gone to my own session code without using php's
built in sessions, I don't have problems at all anymore.
Josh
------------------------------------------------------------------------
[2003-01-08 01:19:06] bug20449 at rehner dot net
My script will fail in as short as 1 request or over 1000 requests.
I did set session.gc_probability = 0 but still fails.
-Ryan
------------------------------------------------------------------------
[2003-01-08 01:13:53] phpBug20449 at rehner dot net
I CAN REPRODUCE THIS BUG!!!
Sort of…
I too had my software working perfectly for over a year under PHP 4.0.6
/ Apache 1.3.24 on Win2000 Pro (development) Win2000 Server
(production).
After upgrading to PHP 4.2.3 / Apache 1.3.27 I've been pulling my hair
out with this session disappearing problem. I tried just about
everything I could find in the bug lists mentioned in this bug to no
avail, short of trying latest CVS. I'm short on time and resources and
from what I read; they haven't fixed the problem yet.
As far as it being a serialization problem I don't think so. None of
my session variables are more complex than a string. So I wrote a
simple test and I know it's not IE specific. I could recreate the
problem with IE 6, Opera 6.05, Netscape 7.0, and Mozilla 1.3
I striped the test to be as simple as 1 integer as a session variable
and it would still die.
I removed the counting aspect to verify it wasn't a write problem, just
reading session data would kill it as well.
The error seems to occur if requests are made to the session variable
from different files close enough together in time to cause one to
"lock" the session file while the next request tries to access it.
Resulting in this error:
Warning: open(c:\tmp\sess_c6bdd642a5d88639e785ec13b0d2f126, O_RDWR)
failed: Permission denied (13) in c:\apache\htdocs\testing\session2.php
on line 10
Obviously it DOES have permission or else my scripts would fail all the
time.
This does fail on all Windows servers I tested with PHP 4.2.3. I did
try it on a Slackware 8.1 with PHP 4.2.1 / Apache 1.3.24 and it did NOT
fail.
I hope this helps track down the elusive bug so I don't have to
downgrade back to 4.0.6 but I'm left with few options.
I will try a CVS if you think it might fix it but I've already spent
more time than I have on this. And from what I read, they don't fix
it. If you need more info just ask. I'll try to do what I can.
My test consists of 4 files. Obviously the comments are not part of
the files.
//----------------file: session0.php --------------------
// Just used to clean things up when they get ugly.
//-------------------------------------------------------
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head><title>Session Test</title></head>
<body>
<pre>
<?php
session_start();
session_unset();
session_destroy();
print_r($_SESSION);
?>
</pre>
<a href="session1.php">Load Values</a>
</body>
</html>
//----------------file: session1.php --------------------
// Sets up the session vars
//-------------------------------------------------------
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head><title>Session Test</title></head>
<body>
<pre>
<?php
session_start();
$_SESSION['count'] = 0;
$_SESSION['test_int'] = 2;
$_SESSION['test_real'] = 3.5;
$_SESSION['test_char'] = 'a';
$_SESSION['test_string'] = 'abc';
$ary = array();
$ary[] = 'z';
$ary[] = 'y';
$ary[] = 'x';
$_SESSION['test_array'] = $ary;
print_r($_SESSION);
?>
</pre>
<a href="session0.php">Reset</a>
<br>
<a href="session2.php">Start Test in single page</a>
<br>
<a href="sessionFrames.php">Start Test in frames</a>
</body>
</html>
//----------------file: session2.php --------------------
// Calls itself repeatedly, increments count
//-------------------------------------------------------
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head><title>Session Test</title></head>
<body>
<pre style="font-size: 8pt;">
<?php
session_start();
if (isset($_SESSION['count'])) {
$_SESSION['count']++;
} else { ?>
<script><!--
alert("Session LOST!");
--></script>
<?php
}
print_r($_SESSION);
?>
</pre>
Running...
<?php if (isset($_SESSION) && isset($_SESSION['count']) &&
($_SESSION['count'] > 0)) { ?>
<script><!--
setTimeout("document.location.href='session2.php'", 0);
--></script>
<?php } else { ?>
Session LOST!
<?php } ?>
</body>
</html>
//-------------file: sessionFrames.php ------------------
// Defines the frameset
//-------------------------------------------------------
<html>
<head><title>Session Test</title></head>
<frameset rows="50,50,50,50" framespacing="0" border="1"
frameborder="1">
<frame name="1" src="session2.php" scrolling="no" noresize
marginwidth="0" marginheight="0">
<frame name="2" src="session2.php" scrolling="no" noresize
marginwidth="0" marginheight="0">
<frame name="3" src="session2.php" scrolling="no" noresize
marginwidth="0" marginheight="0">
<frame name="4" src="session2.php" scrolling="no" noresize
marginwidth="0" marginheight="0">
<noframes>
<p>This page uses frames, but your browser doesn't support
them.</p>
</noframes>
</frameset>
</html>
------------------------------------------------------------------------
The remainder of the comments for this report are too long. To view
the rest of the comments, please view the bug report online at
http://bugs.php.net/20449
--
Edit this bug report at http://bugs.php.net/?id=20449&edit=1
