Re: Re: HTTP doesn't work after upgrading from 2.0.55 to 2.2.15

[Chen Chien-Yu <honercek@xxxxxxxxx>]

Hi William,

I gave you the incorrect information in the previous post.

cgi_read_stdout() in cgi_bucket_read() in mod_cgi.c, the data length is 806 followed by 0. (Not the 0 and 5 buckets)

then ap_core_output_filter() found a 0 length bucket, and appended a last-chunk bucket?

Thanks

Best regards,

honercek

On Fri, May 28, 2010 at 4:02 PM, Chen Chien-Yu <honercek@xxxxxxxxx> wrote:

Hi William,

This is the place I found the 0 and 5 length bucket, cgi_read_stdout() in cgi_bucket_read() in mod_cgi.c which is the function for reading the data from the CGI bucket. So can I say that mod_cgi is the module my Apache uses instead of the mod_cgid. (And I didn't see the mod_cgid through "httpd -l" command)

Is the problem possible in the CGI or mod_cgi, just like the prediction I did in the former post?

Thanks

Best regards,

honercek

On Fri, May 28, 2010 at 12:14 PM, William A. Rowe Jr. <wrowe@xxxxxxxxxxxxx> wrote:

On 5/27/2010 11:04 PM, Chen Chien-Yu wrote:
> Hi William,
>
> Refer to your words, so it's a bug in the bucket brigade mechanism?
> Should I report it to the Apache bug system?
> The two buckets are processed in the ap_core_output_filter(), and in the
> case of EOS bucket (APR_BUCKET_IS_EOS is true)
> That's why, the specific two buckets aren't sent out anymore..?!

There is no bug in the core, when EOS is hit, that's it.

What module creates this missing 5 byte bucket?  Was it a 0 len chunk header?
If that never hits the wire, it sounds like it could be a bug, but I've seen
no other reports of similar buggy behavior in 2.2.15, so I tend to suspect a
module, and most likely an external one.