httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Hartill <r...@imdb.com>
Subject Re: [BUG]: "Truncated pages" on OTHER:SVr4 i386
Date Tue, 04 Mar 1997 14:58:12 GMT
On Tue, 4 Mar 1997, C.T.Nadovich wrote:

> > On Mon, 3 Mar 1997 chris@jtan.com wrote:
> > 
> > > Submitter: chris@jtan.com
> > > Operating system: OTHER:SVr4 i386, version: 4.0.3.6
> > > Version of Apache Used: 1.2b7
> > > Extra Modules used: mod-status
> > > URL exhibiting problem: http://home.jtan.com:8081/bbs
> > > 
> > > Symptoms:
> > > --
> > > Pages are incomplete
> > > read: no more processes in error_log
> >   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> > 
> 
> I've tracked this bug down to the controversial lingering_close()
> routine in http_main(). It seems that the read() in lingering_close() 
> is returning EWOULDBLOCK instead of the acceptable ECONNRESET under
> certain circumstances. 
> 
>     /* Log any errors that occurred (client close or reset is not an error) */
>     
>     if (select_rv < 0)
>         log_unixerr("select", NULL, "lingering_close", r->server);
>     else if (read_rv < 0 && errno != ECONNRESET)
>         log_unixerr("read", NULL, "lingering_close", r->server);
> 
> 
> The strerror() string for EWOULDBLOCK is "No more processes", which is
> very misleading. 
> 
> Are you familiar with the issues embodied in lingering_close() and the
> FIN_WAIT_2 problem exhibited by some browsers?

No, but I know some people who are :-). I'll put you in touch with them
in a moment.

> I think this bug may be
> simply a portability issue on the operation of read() and fcntl() for
> POSIX.1 SVr4 variants. 
> 
> -- 
> Chris Nadovich                                    +1 215 257 8708 (voice)
> http://www.jtan.com/chris                         +1 215 257 8154 (fax)
>                              73 de KD3BJ SK ..


regards,
rob
_______________________________________________________________________
The Oscars(R) are coming.. http://us.imdb.com/Oscars/oscars_1996
			.. no hype, just the facts.


Mime
View raw message