Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 20 Dec 1999 14:14:23 -0800 (PST)
From:      "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net>
To:        msmith@FreeBSD.ORG (Mike Smith)
Cc:        jeffrl@wantabe.com (Jeffrey J. Libman), freebsd-stable@FreeBSD.ORG
Subject:   Re: mountd and rpc.statd won't run
Message-ID:  <199912202214.OAA44113@gndrsh.dnsmgr.net>
In-Reply-To: <199912202157.NAA00717@mass.cdrom.com> from Mike Smith at "Dec 20, 1999 01:57:47 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
> > 
> > after moving 54 virtual hosts to this machine, rpc.statd will not run: it
> > hangs; mountd runs (-r or not -r), but will not serve up mounts and dies.
> 
> Sounds like you may have DNS issues with one or more of the virtual IPs. 
> 
> > 2) configure the loopback interface properly. (i assign all virtual ip's
> > to the lo0 interface).
> 
> That's actually not really a good idea.  Bind them to the interface that 
> they'll be communicating on; that way you don't need to route them.

That one is actually a double edge sword.  Binding them to the ethernet
interface leads to arp table explosion on your routers and/or any
other boxes on the network.  We prefer to route them to lo0 in subnet
chunks, but then we also run ospf on these boxes so that routing is
a non-issue, we can also move a virtual from one server to another very
easily.

Also if you are managing large chunks of virtuals you have to go to
supernetted ethernet IP addresses, and _lots_ of boxes can't cope
with that.

> 
> > 	b) it has been claimed that the virtual ip's need netmask
> > 	0xffffffff rather than 0xffffff00. i have tried it both ways.
> 
> That's also not correct.  The virtual addresses need to be configured 
> _correctly_, where all the normal rules regarding netmask overlap are 
> followed.
> 
> ie.  if you have three 'virtual' addresses 10.0.0.1, 10.0.0.2 and 
> 10.0.0.3, you would add them as:
> 
> 10.0.0.1 / 255.0.0.0
> 10.0.0.2 / 255.255.255.255
> 10.0.0.3 / 255.255.255.255

Actually
10.0.0.1/32
10.0.0.2/32
10.0.0.3/32
works just fine...  and I can have another box that has
10.0.0.4/32
10.0.0.5/32
on it without a bit of problem.  If one uses the /8 you have above
it can create some interesting routing problems for the box that
it lives on if you have 10/8 variablly subnetted and or other
viruals on other boxes in that IP space.

> 
> In other words, you add an aliased address with the correct netmask, 
> unless it overlaps a network previously established by another address, 
> in which case you add it with an all-ones netmask.

We _always_ use /32 (all ones) and aggregate a route if a box has a
large enough chunk of virtuals to be able to create a /30 or larger.


-- 
Rod Grimes - KD7CAX @ CN85sl - (RWG25)               rgrimes@gndrsh.dnsmgr.net


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199912202214.OAA44113>