From: Tim Kennedy <sugarat@thunderhold.sugarat.net>
On Wed, 21 Jul 1999, Berg, Ivan wrote:
> 1. Sending a ping to the broadcast address of several subnets results in
> DUP packets from almost every responding hosts. The router of these subnets
> has directed broadcasts turned on. Other routers have directed broadcasts
> turned off and thus I can not reach them from where I am at. The man page
> for ping states this:
> DUPLICATE AND DAMAGED PACKETS
> Ping will report duplicate and damaged packets. Duplicate packets
> should
> never occur, and seem to be caused by inappropriate link-level retrans-
> missions. Duplicates may occur in many situations and are rarely (if
> ev-
> er) a good sign, although the presence of low levels of duplicates may
> not always be cause for alarm.
>
> Now, this does not sound good, could this be normal?? cause for alarm?? what
> could be causing it.
I suspect that pinging a broadcast address is different situation.
You send out a single ECHO. You receive the first ECHO-REPY. Good.
You receive another ECHO-REPLY to the same ECHO. Huh? DUPLICATE.
That is normal behavior, when pinging a broadcast address. Every node on
that netwok will respond. If you ping a node that is not a broadcast
address and you receive DUP packets, then yes, that _is_ bad.
> 2. I have read networking books that state when using a classfull routing
> protocol such as RIP that you should not use masks other than the standard
> class ones. They say you could possibly do it but that the mask should
> always be larger than what you are using by default on other equipment. For
> instance, if you have a class C subnet with a mask of 24, then it states
> that you should only use masks > 24. e.g. DONT use a 23 bit net mask. My
> question is: why is this bad?? It doesn't seem to cause problems. For
> instance, on network using RIP, we want subnet with 512 hosts(23 bit
> netmask), since all hosts don't have netmask set properly, two routes are
> used. For example, 192.168.1.0/23 would be the proper way, however, you can
> also use 192.168.1.0/24 and 192.168.2.0/24 since RIP doesn't know about 24
> bit netmask and this will also work. Of course, you lose 192.168.2.0 and
> 192.168.1.255 as valid addresses, but it seems to work.
>
> Why is this bad??
Possible has to do with route aggregation/summarization which RIP doesn't
do. The smaller the netmask you use to announce routes with RIP, the more
routes you have to announce, the larger your RIP table is. Using a routing
protocol like OSPF that can aggregate your samller routes, this becomes
less of an issue.
We actually use rip to announce individual ip addresses (/32) that are
dialed into our network, then have the routers perform RIP->OSPF
translation and summarize the routes outbound.
Just a guess.
-Tim
Timothy Kennedy
sugarat.net, Network Management Resources
http://www.sugarat.net
--------------------------- ONElist Sponsor ----------------------------
ONElist members are using Shared Files in great ways!
http://www.onelist.com
Are you? If not, see our homepage for details.
------------------------------------------------------------------------
This archive was generated by hypermail 2b29 : Mon Mar 06 2000 - 19:00:58 PST