TROUBLESHOOTING CAMPUS NETWORKS

by Priscilla Oppenheimer and Joseph Bardwell

Updates

While we made every effort to ensure that the information in Troubleshooting Campus Networks is timely and accurate, minor errors do creep in during the production of a 600-page book. This Web page will keep you posted on any errors that we find and include updated information about troubleshooting campus networks.

Chapter 3: Troubleshooting and Analyzing Ethernet Networks

On page 75, in Figure 3.4, in the Novell raw frame format, in the bottom-right corner, DST should say FCS. The final field in all varieties of the Ethernet frame is the Frame Check Sequence (FCS), not Destination (DST). DST is the first field in all varieties, following the Preamble (P).

The following figure shows the frame formats.

Note that the figure shows the industry-standard names for the frames. According to IEEE, an Ethernet 802.3 frame always includes an 802.2 LLC header so there's no need to have different names for 802.3 and 802.2 frame formats. Novell, on the other hand, supported a frame type that used only 802.3. For that reason, Novell uses slightly different terminology. The following table maps industry-standard names to Novell names.

 Industry-Standard Name Novell Name 
 Ethernet II or Ethernet Version 2  Ethernet_II
 IEEE 802.3 (with 802.2)  Ethernet_802.2
 IEEE SNAP  Ethernet_SNAP
 Novell Raw  Ethernet_802.3


On page 84, we discuss Ethernet padding which makes an Ethernet frame adhere to the IEEE 802.3 requirement that a frame must be at least 64 bytes, counting the header and FCS. The padding should be filled with zeros, per RFC 1042, but we have seen many cases where it is filled with readable text. In the book, we mention the worst case we saw, which occurred when a user logged into a database application. The user's machine sent an encrypted password in the login frame, but the next frame from the user needed padding, and that frame included the unencrypted password in the Ethernet padding!

We had heard that a bug fix from the database vendor fixed the problem. By cleaning up the "dirty buffer" that held the password, the problem could be avoided. In 2003, CERT published a Vulnerability Note related to Ethernet drivers that don't pad with all zeros. Based on the information in this Note, we suspect that the problem we saw was related to the Ethernet driver and not just to the database application.

Chapter 4: Troubleshooting and Analyzing Wireless Networks

On page 126, below the first "Assume," the page formatter inserted an extra "p."

log 0.1 = -1.0 (because 10p-1 = 0.1)

should read

log 0.1 = -1.0 (because 10-1 = 0.1)


Also on page 126, in the final line of that same example, the decimal point is in the wrong place.

10 log (P1/P2) = 10 * -0.107905 = -10.7905 dB

should read

10 log (P1/P2) = 10 * -0.107905 = -1.07905 dB


On page 150, "15 square feet" and "50 square feet" should read "15 feet squared" and "50 feet squared."

Chapter 5: Troubleshooting and Analyzing the Spanning Tree Protocol

On page 168, the default value for Forward Delay is incorrectly stated as 20 seconds. This should say 15 seconds. Forward Delay is discussed throughout the chapter and the other references correctly say 15 seconds.

Chapter 5: Troubleshooting and Analyzing the Spanning Tree Protocol

On page 185, the page formatter skewed the letters below the output of the debug span events command.

The final two lines on that page should look like this:

0000000000800000021676324200000077800000100D95840080420200140002000F00
A   B C D E   F           G       H   I           J K L   M   N   O

Chapter 6: Troubleshooting and Analyzing Virtual LANs

On page 205, the statement that reads "A workstation configured as a TFTP server can act as a VMPS server" should read "A workstation configured as a TFTP server stores the VMPS database." A high-end Cisco switch downloads the database and acts as the server. There's more information about VMPS at this Cisco Web page:

Configuring Dynamic Port VLAN Membership with VMPS

Chapter 7: Troubleshooting and Analyzing Campus IP Networks

On page 242, in Figure 7.2, the second Total Length should say Fragment Offset, as shown below.

Chapter 8: Troubleshooting and Analyzing Campus IP Routing Protocols

On page 297, the word "not" is missing from this sentence: "Classful routing protocols also do support a discontiguous subnet." It should read: "Classful routing protocols also do not support a discontiguous subnet."

Chapter 9: Troubleshooting and Analyzing TCP, UDP, and Upper-Layer IP Protocols

On page 369, in Table 9.5, the page formatter inserted an X in the port number for NetWare Core Protocol (NCP) when NCP is used with TCP or UDP. The port number should simply say 524, not 5x24.


Back to the Troubleshooting Campus Networks home page.

Copyright © Priscilla Oppenheimer.

Hosted by Open Door Networks.