Re: Error Message from my proxy server

Marius Tomaschewski <[email protected]> Tue, 30 Jul 2002 17:24:33 +0200
Newsgroups gmane.linux.suse.proxy-suite
Organization SuSE Labs, Product Developement
Message-ID <[email protected]>
On Tue, Jul 30, 2002 at 10:28:09AM -0400, Ruiyuan Jiang wrote:
> Hi, Marius
> 
> Here is the whole log since yesterday I changed in the
> configuration from "LogLevel WRN" to LogLevel DBG" and
> restarted ftp proxy server daemon. The messages are the
> same as LogLevel WRN and DBG. Thanks.

Hmm...

The FIONREAD ioctl fails without to set any errno value...
(the syslog_error function adds it to the log message if set).

It is is redefined on solaris (see common/com-syslog.c) to
use the streams I_NREAD ioctl instead of FIONREAD.


Please take a look in "man iostream" (if I remember right);
perhaps they have changed the behaviour of I_NREAD ??

The question is what this means on solaris9 that you are
using... on solaris7 & 8 this does not happen, except on
real socket errors.


It seems to me, solaris9 returns -1 on EOF?!
Or do you have noticed any transfer/connection aborts
when this message happens?

In the EOF case you can ignore the error, i.e. using
(additionaly to the patch):

syslog_write(T_DBG, "can't get num of bytes: %s %d=%s",

instead of the syslog_error call, so the socket simply
gets closed.

The patch I've sent you last time closes the socket if
FIONREAD/I_NREAD fails, but still displays it as error:

if( (cnt=ioctl(hls->sock, FIONREAD, &len)) < 0) {
	hls->ernr = errno;
	syslog_error("can't get num of bytes: %s %d=%s",
	     hls->ctyp, hls->sock, hls->peer);
	close(hls->sock);
	hls->sock = -1;
	return;
}

Kind regards,
 Marius Tomaschewski <[email protected]>
--
 SuSE Linux AG, NÃŒrnberg - SuSE Labs, Product Developement
 PGP public key available:   http://www.suse.de/~mt/mt.pgp
 Fprint:  EA 1F 92 75 1A F9 82 07  A1 28 DE 7A 32 E8 97 18

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]