|
Stephen Rowles wrote:
> >This doesn't sound like an LVS problem straight off. What happens if you
> >remove the real-server from the LVS and try to mount the home directories
> >onto the real-server box?
>
> I don't think it is specifically an LVS problem - more likely a NAT / NFS
> problem. The machines work fine when connected directly to network (they
is this a "yes" answer to the above question?
> are the same install as is used for other standard linux boxes in the lab,
> but with some LVS / NAT modifications)
>
> I think that the problem might lie more with the NAT side of things due to
> the fact that the NFS server thinks the machine is "<unknown>" rather than
> a recognised host despite being in all the appropriate host yp files.
The only NAT related problem I can think of would stop everything you are
doing from working.
In the real-server setup, you will have run on the director a command like
$ipchains -A forward -p tcp -j MASQ -s RIP telnet -d 0.0.0.0
you should have a separate line for each RIP here (eg 192.168.1.10)
and not one line with the network (eg 192.168.1.0)
On the real-server you'll have the default gw as the director.
If you are mounting disks from backend machines, you have to make sure that
the default gw isn't stopping packets between the backend machines and
real-servers.
(connect to the backend machines via a different network or host routes, using
a second IP on the real-server)
Also don't mount the disks onto the director, in LVS it's just a switch and
offering no services. Mounting disks onto the director will stop the director
from routing requests to those ports.
> I realise that this is slightly off-topic as there probably isn't an LVS
> problem, but I thought it might be something other people had experienced
> or a problem that others might have in the future.
We need to get LVS to work in the context it's being used, so this is
a good place to figure out this problem :-)
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
|
