Mailing List Archive

On Thu, 17 Aug 1995, Andrew Wilson wrote:
> 193.131.197.60 - - [17/Aug/1995:10:33:26 +0100] "NULL" 200 864

Port 80 of telescope.org is still with 0.8.7 and gets lines like...

ix-atl11-01.ix.netcom.com - - [15/Aug/1995:05:37:44 +0100] "(null)" 200 -
slc100.xmission.com - - [15/Aug/1995:06:45:02 +0100] "(null)" 404 837

The first line occurs a few times (325 times out of 336k in current log)
the second is confusing - how can a NULL request return a 404 not found?

Mark
Mark J Cox, mark@telescope.org -- URL:http://www.telescope.org/mark.html
University of Bradford, England ---------- tel +44.1274.384070/fax 391333

>Florent Guillaume wrote:
>>
>> > On Thu, 17 Aug 1995, Andrew Wilson wrote:
>> > > 193.131.197.60 - - [17/Aug/1995:10:33:26 +0100] "NULL" 200 864
>>
>> >[Mark]
>> > ix-atl11-01.ix.netcom.com - - [15/Aug/1995:05:37:44 +0100] "(null)" 200 -
>> > slc100.xmission.com - - [15/Aug/1995:06:45:02 +0100] "(null)" 404 837

Jim replied:
>
>There's no patch needed, since the patch that _creates_ the "NULL"
>string in the log _is_ the patch.

Yes, a patch is required. All three of the above log entries are
deceptive. They should not record 200 unless the request was accepted
and a response returned. Request Timeouts should record 408 (if it
really is a timeout) or 400 (if the request is just invalid).

.......Roy

> Yes, a patch is required. All three of the above log entries are
> deceptive. They should not record 200 unless the request was accepted
> and a response returned. Request Timeouts should record 408 (if it
> really is a timeout) or 400 (if the request is just invalid).
>
>I can see that this ought to be fixed, but I'm not at all sure it should
>hold up all the other bug fixes which we haven't distributed yet.

Ooops, sorry, I meant eventually.

+1 on the current release.

......Roy

Yes, a patch is required. All three of the above log entries are
deceptive. They should not record 200 unless the request was accepted
and a response returned. Request Timeouts should record 408 (if it
really is a timeout) or 400 (if the request is just invalid).

I can see that this ought to be fixed, but I'm not at all sure it should
hold up all the other bug fixes which we haven't distributed yet. (It's
easy enough to do... toss in

if (timeout_req->status == 200) timeout_req->status = 408;

at the obvious point in http_main.c... I'm planning to try to mop up a
number of things like this which are hanging, having been reported but
with no fix, on the weekend, and then (in response to all the DEMOCRACY!
DEMOCRACY! hoo-hah) put that with other bug-fixes received on a ballot...).

rst

> On Thu, 17 Aug 1995, Andrew Wilson wrote:
> > 193.131.197.60 - - [17/Aug/1995:10:33:26 +0100] "NULL" 200 864

>[Mark]
> ix-atl11-01.ix.netcom.com - - [15/Aug/1995:05:37:44 +0100] "(null)" 200 -
> slc100.xmission.com - - [15/Aug/1995:06:45:02 +0100] "(null)" 404 837


This happens when the server does a 'read timed out' (check the
error_log at the same date and time). In fact the server hasn't yet
received the first line of the request, so the "the_request" field isn't
initialized yet.

I'll let someone who knows this part of the code better than me send a
patch.


Florent