# "zero size buf in output" flooding the error.log

**URL:** <https://www.ruby-forum.com/t/zero-size-buf-in-output-flooding-the-error-log/212792>\
**Category:** NGINX\
**Created:** [October 14, 2011, 3:46am UTC](https://www.ruby-forum.com/t/zero-size-buf-in-output-flooding-the-error-log/212792 "2011-10-14T03:46:37Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![dubstep](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/d/cdc98d/32.png) [@dubstep](https://www.ruby-forum.com/u/dubstep)\
**Post date:** [October 14, 2011, 3:46am UTC](https://www.ruby-forum.com/t/zero-size-buf-in-output-flooding-the-error-log/212792/1 "2011-10-14T03:46:37Z")

</div>

I am using the external NGiNX\_HTTP\_Push\_Module, and whenever a  
response is issued by one of its channels, I get this error logged:

2011/10/12 19:33:30 [alert] 10808#0: \*64 zero size buf in output t:1  
r:0 f:0 0000000001240C98 0000000001240C98-0000000001240C98  
0000000000000000 0-0, client: 127.0.0.1, server: localhost, request:  
“GET /event?id=hJsrnGIdbngARLNBiNMoIO HTTP/1.1”, host: “XXXXXXXXXXX”,  
referrer: “XXXXXXXXXXXXXXXXXXX”

Those ‘client’ and ‘server’ fields are strange, because they always  
point to localhost, even if connection is external.

By “greping” through the source I found the message is logged by  
ngx\_output\_chain.c:100, where I saw that this is clearly a debug  
message, because of all the pointers it prints. So I thought it might  
be a bug on the NGiNX\_HTTP\_Push\_Module that triggers the situation.  
What might be the issue on the module? What can trigger this message?

–  
Lucas Clemente Vella  
removed\_email\_address@domain.invalid