Chapter 4

Chapter 4

4.1

In the commands we typed to generate the output shown in Figure 4.4, what would happen if, after verifying that the local ARP cache was empty, we type the command

 bsdi %  rsh svr4 arp -a  

to verify that the ARP cache is also empty on the destination host? (This command causes the arp -a command to be executed on the host svr4 .)

A:

Issuing an rsh command establishes a TCP connection with the other host. Doing that causes IP datagrams to be exchanged between the two hosts . This requires the ARP cache on the other host to have an entry for our host. Therefore, even if the ARP cache was empty before we executed the rsh command, it's guaranteed to have an entry for our host when the rsh server executes the arp command .

4.2

Describe a test to determine if a given host handles a received gratuitous ARP request correctly.

A:

Make sure that your host does not have an entry in its ARP cache for some other host on its Ethernet, say foo. Make sure foo sends a gratuitous ARP request when it bootstraps, perhaps running tcpdump on another host when foo bootstraps. Then shut down the host foo and enter an incorrect entry into the ARP cache on your system for foo, using the arp command and being sure to specify the temp option. Bootstrap foo and when it's up, look at your host's ARP cache entry for it to see whether the incorrect entry has been corrected .

4.3

Step 7 in Section 4.2 can take a while ( milliseconds ) because a packet is sent and ARP then waits for the response. How do you think ARP handles multiple datagrams that arrive from IP for the same destination address during this period?

A:

Read Section 2.3.2.2 of the Host Requirements RFC and Section 11.9 of this text .

4.4

At the end of Section 4.5 we mentioned that the Host Requirements RFC and Berkeley-derived implementations differ in their handling of the timeout of an active ARP entry. What happens if we're on a Berkeley-derived client and keep trying to contact a server host that's been taken down to replace its Ethernet board? Does this change if the server issues a gratuitous ARP when it bootstraps?

A:

Assuming that a completed ARP entry existed for the server on the client when the server was taken down, if we continually try to contact the (down) server, the ARP timeout gets extended for another 20 minutes. When the server finally reboots with a new hardware address, if it doesn't issue a gratuitous ARP, the old, invalid ARP entry will still exist on the client. We won't be able to contact the server at its new hardware address until we either manually delete the ARP cache entry or stop trying to contact it for 20 minutes.



TCP.IP Illustrated, Volume 1. The Protocols
TCP/IP Illustrated, Vol. 1: The Protocols (Addison-Wesley Professional Computing Series)
ISBN: 0201633469
EAN: 2147483647
Year: 1993
Pages: 378

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net