Helmuth Gronewold
2014-07-02 14:32:34 UTC
Dear list,
I really need your help with this. I have two webserver clusters with
mod_suphp 0.7.1 and Apache 2.2 (squeeze-lts). BalanceNG is placed
infront of the webservers to load balance user requests in DSR mode. PHP
sessions and data is stored on a reliable NFS server.
Multiple times a day users get an internal server error and the Apache
logs say:
[Wed Jul 02 09:34:32 2014] [error] [client 9x.xx.xx.64] Handler for
(null) returned invalid result code 70014, referer:
http://www.example.com/user_update.php?action=upload&id=292931
In more than two cases the error was reproducible for the user only, but
was gone the next day or after a few hours. Two months ago 6 out of 8
users from one office got this problem and again it fixed itself the
other day.
As you can see, the referer URL above says action=upload. The problem is
not bound to file uploads. Sometimes scripts that could be used to
upload files only changed two values with a few byte in POST size.
It is also not bound to a specific script. I've seen that error with
multiple different scripts which go from mostly flat PHP files with
structural programming to MVC and OOP. The only thing that most of the
scripts have in common is, that they provide file upload functionality
(which is not used every time...).
I've read every website on the six result pages google returned to this
topic. No one found a clear solution to this. Some say a proxy or
firewall is causing this (pfSense and BalanceNG is used here, but there
are equal rules for all users and no content modification), some say
it's specific to the executed PHP code.
In my opinion the most interesting results are:
1.
http://serverfault.com/questions/231331/handler-for-null-returned-invalid-result-code-70007-causing-error-500
2. http://www.webhostingtalk.com/showthread.php?t=1351337
The first one describes what the error means. I've checked out both the
Apache and suPHP codebase and found out that ap_run_handler() returns
this error. The function is implemented by the handler which in turn,
AFAIK, is mod_suphp.
The second ones poster did some nice debugging and found out the
following:
1. mod_security is not an issue (even though I'm not using
mod_security...)
2. POST requests with forms are involved 100% of the time
3. suphp is involved
4. the PHP script itself is not even executed
5. "this error can also be logged if a form's content-size value in the
POST headers does not match the actual content size"
On the other hand I did not get a "Connection Reset" like he posted.
In my understanding the error is generated when there is a communication
problem between Apache and suPHP. But I'm really not sure about that.
Every single advice to track down this problem would be great!
Thanks in advance,
Helmuth
I really need your help with this. I have two webserver clusters with
mod_suphp 0.7.1 and Apache 2.2 (squeeze-lts). BalanceNG is placed
infront of the webservers to load balance user requests in DSR mode. PHP
sessions and data is stored on a reliable NFS server.
Multiple times a day users get an internal server error and the Apache
logs say:
[Wed Jul 02 09:34:32 2014] [error] [client 9x.xx.xx.64] Handler for
(null) returned invalid result code 70014, referer:
http://www.example.com/user_update.php?action=upload&id=292931
In more than two cases the error was reproducible for the user only, but
was gone the next day or after a few hours. Two months ago 6 out of 8
users from one office got this problem and again it fixed itself the
other day.
As you can see, the referer URL above says action=upload. The problem is
not bound to file uploads. Sometimes scripts that could be used to
upload files only changed two values with a few byte in POST size.
It is also not bound to a specific script. I've seen that error with
multiple different scripts which go from mostly flat PHP files with
structural programming to MVC and OOP. The only thing that most of the
scripts have in common is, that they provide file upload functionality
(which is not used every time...).
I've read every website on the six result pages google returned to this
topic. No one found a clear solution to this. Some say a proxy or
firewall is causing this (pfSense and BalanceNG is used here, but there
are equal rules for all users and no content modification), some say
it's specific to the executed PHP code.
In my opinion the most interesting results are:
1.
http://serverfault.com/questions/231331/handler-for-null-returned-invalid-result-code-70007-causing-error-500
2. http://www.webhostingtalk.com/showthread.php?t=1351337
The first one describes what the error means. I've checked out both the
Apache and suPHP codebase and found out that ap_run_handler() returns
this error. The function is implemented by the handler which in turn,
AFAIK, is mod_suphp.
The second ones poster did some nice debugging and found out the
following:
1. mod_security is not an issue (even though I'm not using
mod_security...)
2. POST requests with forms are involved 100% of the time
3. suphp is involved
4. the PHP script itself is not even executed
5. "this error can also be logged if a form's content-size value in the
POST headers does not match the actual content size"
On the other hand I did not get a "Connection Reset" like he posted.
In my understanding the error is generated when there is a communication
problem between Apache and suPHP. But I'm really not sure about that.
Every single advice to track down this problem would be great!
Thanks in advance,
Helmuth