LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

RE: [lvs-users] LVS-DR generates TONS of icmp unreachables

To: "Joseph Mack" <mack@xxxxxxxxxxx>
Subject: RE: [lvs-users] LVS-DR generates TONS of icmp unreachables
Cc: <wensong@xxxxxxxxxxxx>, "Jerry Glomph Black" <black@xxxxxxxx>, "Julian Anastasov" <uli@xxxxxxxxxxxxxxxxxxxxxx>, <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
From: "Jivko Velev" <jiko@xxxxxxxxxx>
Date: Thu, 20 Jan 2000 13:40:30 -0800
> > > In VS-DR the director never sees the replies from the
> realservers and will
> > > have no way of knowing which realserver is responsible for these ICMP
> > > replies. Not seeing the replies is a Good Thing in that it keeps the
> > > throughput of the LVS high. However it does make monitoring
> the health of
> > > the LVS difficult. The problem of PORT_UNREACH is discussed
> in sect 14.16
> > > of the HOWTO in this context.
> >
> > TCP connection is a full duplex connection and as i understood
> with VS-DR
> > and VS-TUN Redirector doesnt see only the server-client side of
> the channel,
> > but the other side we have TCP confirmation packets flowing
> from client to
> > sever and they still go throught the redirector.
> > Is that true ?
>
> yes. The director doesn't see the packets going from the realserver to the
> client, but does see the packets going from the client to the realserver.
>
> Joe
>
So, all confirmation packets of one TCP connection are redirected
/persitently/ from VIP to RIP /because it wont work otherwise/.
My suggestion to the designers & implementators of LVS is to redirect the
ICMP packets in the same way /all required info is in ICMP packet/
and everything will be OK.

Jiko

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