Problems creating small subnet between servers
2007-12-25 8:06:00
Here's my original posting:
> I have another rather serious problem that I hope sun-managers can
> help me with. I have the following environment:
>
> 4/470 running 4.1.3
> 670MP running 4.1.3
>
> These two machines are the main engineering servers. On the eng ethernet
> there's appx 50 sparcs as well as 25 or so xterminals. The two servers
> run a source control package called ClearCase. The two servers rely
> heavily upon each other.
>
> The problem I had/have is that there's ALOT of nfs traffic being
> generated between these two servers. I decided to build a small
> subnet between these two machines, and funnel all the traffic
> between them onto this small subnet, thus easing the load that's
> currently on the main ethernet. Here's how it looks:
>
>
>
> small subnet
> -----------------
> | |
> (lenny) 4/470 670MP (fats)
> | |
> ----------------------------------------------------------------
> main engineering subnet
>
>
> The way I have things set up is the ie1 on lenny (to the small subnet)
> is called "lenny-gw". Likewise, the le1 on fats is called "fats-gw".
> Each machine has a default route to the router connected to the main
> ethernet:
>
> /usr/etc/route add default 128.185.123.2 1
>
> After the default route is enabled in the /etc/rc.local, I added
> an additional line to each to specifically route the hosts together:
>
> (on fats):
>
> /usr/etc/route add lenny.proteon.com fats-gw.proteon.com 1
>
> (on lenny):
>
> /usr/etc/route add fats.proteon.com lenny-gw.proteon.com 1
>
>
> The way I understand these lines is "add a route to lenny, have it
> go thru gateway fats-gw, and it's one hop". Am I interpreting this
> correctly?
>
> Anyways, When I rebooted both machines, I could tell things weren't
> really correct (I got some automounts not mounting). When I try and
> ping from either machine to the other I get these errors:
>
> (from fats to lenny):
>
> PING lenny.proteon.com: 56 data bytes
> ICMP Time Exceeded from localhost (127.0.0.1)
> for fats-gw.proteon.com (128.185.122.17) to lenny.proteon.com (128.185.123.120) port 8170
> ICMP Time Exceeded from localhost (127.0.0.1)
> for fats-gw.proteon.com (128.185.122.17) to lenny.proteon.com (128.185.123.120) port 51468
> ICMP Time Exceeded from localhost (127.0.0.1)
> for fats-gw.proteon.com (128.185.122.17) to lenny.proteon.com (128.185.123.120) port 30294
>
>
> (from lenny to fats):
>
> PING fats.proteon.com: 56 data bytes
> ICMP Time Exceeded from localhost (127.0.0.1)
> for lenny-gw.proteon.com (128.185.122.120) to fats.proteon.com (128.185.123.17) port 16204
> ICMP Time Exceeded from localhost (127.0.0.1)
> for lenny-gw.proteon.com (128.185.122.120) to fats.proteon.com (128.185.123.17) port 1946
> ICMP Time Exceeded from localhost (127.0.0.1)
> for lenny-gw.proteon.com (128.185.122.120) to fats.proteon.com (128.185.123.17) port 47464
>
>
>
> BIG QUESTION: WHAT'S WRONG WITH THIS? To me it appears everything
> should work. The default route will route everything.
> And the hard mount fats<-->lenny will route the
> traffic between the two via their le1/ie1 ports.
> But it's not working.
>
>
> The alternative that people suggested I do is to make custom YP maps
> for lenny and fats and use the "automount -f -m /etc/auto.master"
> deal on each machine. I don't want to do this because 1) it's more
> variables to deal with when it comes to the upkeep of the maps, and
> 2) the routes SHOULD WORK.
>
>
> help please.... I will summarize... and hopefully with the info that
> lead me to the sucess of this project!
>
> Thanks!
>
> -Gary
>
My problem was that I was using "1" as the metric, when I should have
used "0" (since I was on the same physical piece of wire, there were
no hops). Once I changed 1 to 0, things worked just dandy.
My thanks to:
ems@ccrl.nj.nec.com (Ed Strong)
lhume@macadam.mpce.mq.edu.au (Leigh Hume)
mandrews@hq.videocart.com (Mike Andrews)
Eckhard.Rueggeberg@ts.go.dlr.de
cyerkes@jpmorgan.com (Chuck Yerkes)
pjw@sma.usna.navy.MIL (Peter J. Welcher -- math FACULTY
Patrick O'Callaghan <poc@usb.ve>
Christian Lawrence <cal@soac.bellcore.com>
lmdrsm@lmd.ericsson.se (LMD/T/AM Roar Smith)
stern@sunne.East.Sun.COM (Hal Stern - NE Area Systems Engineer)
Mike Raffety <miker@il.us.swissbank.com>
Thanks all!
-Gary
Comments
Got something to say?
You must be logged in to post a comment.
