multihomed jumpstart

Heinz Müller

unread,

Oct 10, 2008, 1:18:37 PM10/10/08

to

I've been looking for this answer for quite some time, and I found it
somewhere on the web... and it worked.

Having a Solaris 10 (6/06) Jumpstart server (Not JET) with two NICs on
separate networks my not jump from both nets.

I could not get my clients on one network to jump. Only the clients on
the jumpstart server's primary network would jump correctly.

All I had to do was add another entry in my /etc/defaultrouter file on
the jump server.

cat /etc/defaultrouter
192.168.10.1
192.168.11.1

You could also do a route add default statement on the fly.

No need to have a /etc/notrouter file any more on Solaris 10, see "man
notrouter". By default it will not route between the NICs.

Darren Dunham

unread,

Oct 10, 2008, 8:13:12 PM10/10/08

to

"Heinz M?ller" <onkel...@mscologne.de> wrote:
> I've been looking for this answer for quite some time, and I found it
> somewhere on the web... and it worked.
>
> Having a Solaris 10 (6/06) Jumpstart server (Not JET) with two NICs on
> separate networks my not jump from both nets.
>
> I could not get my clients on one network to jump. Only the clients on
> the jumpstart server's primary network would jump correctly.
>
> All I had to do was add another entry in my /etc/defaultrouter file on
> the jump server.
>
> cat /etc/defaultrouter
> 192.168.10.1
> 192.168.11.1

Was this multihomed server just a boot server, or was it the full
jumpstart server? If the client didn't have to talk to other subnets, I
assume having a bad defaultrouter wouldn't have mattered.

This is a limitation/bug associated with rpc.bootparamd. As the client
begins setting up more networking stuff during boot, it sets its
defaultrouter to be that of the boot server. For a multi-homed boot
server with a single default router (which would be the normal way to do
it), this causes problems for clients not on the default router's
subnet if necessary resources are off-subnet (like the actual jumpstart
server).

--
Darren

Heinz Müller

unread,

Oct 11, 2008, 5:14:23 PM10/11/08

to

"Darren Dunham" <ddu...@taos.com> schrieb im Newsbeitrag
news:scSHk.3223$c45...@nlpi065.nbdc.sbc.com...

> Was this multihomed server just a boot server, or was it the full
> jumpstart server? If the client didn't have to talk to other subnets, I
> assume having a bad defaultrouter wouldn't have mattered.
>
> This is a limitation/bug associated with rpc.bootparamd. As the client
> begins setting up more networking stuff during boot, it sets its
> defaultrouter to be that of the boot server. For a multi-homed boot
> server with a single default router (which would be the normal way to do
> it), this causes problems for clients not on the default router's
> subnet if necessary resources are off-subnet (like the actual jumpstart
> server).
>
> --
> Darren

Darren,

Thanks for you answer, but it was a mistake of mine. I didn't want to post
it to
this newsgroup but to another email adress!

I copied that from another website.

SORRY!!!!

BTW:
What I want to build up is a full jumpstart server mutlihomed to 2 subnets.
If that doesn't work an altervative would be to setup a full jumpstart
server
_and_ a boot server for the second subnet.

Heinz