LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: It looks good, it smells good, but...

To: Joseph Mack <mack.joseph@xxxxxxx>
Subject: Re: It looks good, it smells good, but...
Cc: Joseph Mack <mack.joseph@xxxxxxxxxxxxxxx>, Julian Anastasov <ja@xxxxxx>, lvs-users@xxxxxxxxxxxxxxxxxxxxxx
From: David Rodrigues <drodrig@xxxxxxxxxxxxxxxx>
Date: Wed, 13 Sep 2000 10:41:10 -0400 (EDT)
I'm probably grasping at straws, but...
Do you think the fact that the ethernet cards are running in promiscuous
mode would affect anything? I checked my other LVS server (working) and it
does NOT have the ethernet cards in promiscous mode. Does anyone know how
to shut off promiscuous mode? I'm using an onboard EtherExpress Pro 10/100
card. The driver is in the kernel, not modular.

One other thing. The director box is a Cobalt RaQ 4r. 

On Wed, 13 Sep 2000, Joseph Mack wrote:

> David Rodrigues wrote:
> > 
> > On Wed, 13 Sep 2000, Joseph Mack wrote:
> > 
> > > David Rodrigues wrote:
> > > >
> > > > When I attempt to point my browser at 192.168.1.214, I get the 
> > > > director's
> > > > web page. If I disable httpd on the director, I get the 'unable to 
> > > > connect
> > > > to ...' error message.
> > >
> > > your ipvsadm table looks OK. This means then that the real-server is not
> > > listening on 192.168.1.214:80
> > >
> > 
> > This may be true, but even so the director shouldn't serve the web page
> > itself, should it? I would think I should get a timeout, even if httpd is
> > running on the director. I don't have the local-node feature set up.
> 
> when it works, it only works one way. When LVS is not working there are
> many ways it doesn't work and all bets are off as to what it should or 
> shouldn't
> do. Yours is in a mode where the director is accepting the packets rather 
> than forwarding them. 
> 
> your test (just below) shows that the real-server is serving VIP:80, so my
> answer above doesn't apply (and I'm stumped). 
> 
> from here, unless someone has seen these exact symptoms before, it's brute
> force.
> You could try tcpdump on the client, director and real-server.
> 
> You could try mt script on the LVS webpage. 
> It sets up a 1 director LVS with lots of checks for
> all the ways I got it to fail
> 
> > > can you pull one real-server out of the LVS network, turn off the hidden 
> > > on the
> > > VIP and connect
> > > to VIP:80 on the pulled real-server from another machine?
> > 
> > Yes.
> 
> Joe
> -- 
> Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
> contractor to the National Environmental Supercomputer Center, 
> mailto:mack.joseph@xxxxxxx ph# 919-541-0007, RTP, NC, USA
> 
> 
> 



<Prev in Thread] Current Thread [Next in Thread>