httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: Apache 2.0 on again
Date Sun, 21 Jan 2001 14:33:42 GMT

In a message dated 01-01-21 11:23:58 EST, Ryan writes...

> I just checked the error logs for 2.0.  There were 3010 requests since
> last night, and the only errors were file not found and one malformed
> request.  This is looking very good right now.

The malformed request was mine.

It was a telent request 
GET / HTTP/1.1 (crlf)
with no 'Host:' field and Server correctly
returned 'Malformed request'.

Same request changed to HTTP/1.0 
worked AOK. No 'Host:' field required and home page flew back.

I thought I heard some noise earlier on about
this being an issue but it certainly isn't one now.
Server does what it should.

The 'connect/disconnect' testing won't show in the logs because
Apache won't log anything if the connection drops before
any data has been received.

This is why load-balancing 'polling' doesn't show in
Apache logs. When a load balancer thinks a copy of
Apache has 'gone away' it puts it in the 'panic' list and
starts hammering it with connect/disconnect requests
until it 'connects' OK again and then it pulls it out of
the 'panic' list. Apache 2.0 will obviously deal with this
AOK. Most load balancers all do this 'panic' thing.

Kevin Kiley
CTO, Remote Communications, Inc.

View raw message