Who should attend: All UNIX system administrators, beginning to intermediate network administrators, and advanced users who have to diagnose and fix problems in their TCP/IP networks, or work with others to do so. You should have some familiarity with general networking concepts.

What you will learn: How to isolate, diagnose, and correct the most common failures in TCP/IP networks using a diagnostic decision logic table (DDLT).

The network is down: How do you fix it? This is no time to read a book, and even the best organized textbook is not a good reference in a crisis.

Most sites have no specialist to call when the network breaks, so UNIX system administrators and users are often called on to fix things, but they seldom have a coherent plan of attack. Even people with a great deal of experience with TCP/IP at the application level often lack experience with network pathology.

This tutorial examines a practical problem-solving method using a diagnostic decision logic table (DDLT), developed by the instructor and based on many years' experience. The DDLT gives step-by-step instructions on what to look at, what it should look like, and how to fix it. Even if it's not your job to fix it, you can help get things back on the air quickly by knowing which vital statistics to gather to give the diagnosis a head start.

The tutorial focuses on specific examples of network failures, and introduces and amplifies on TCP/IP and Ethernet concepts. Examples are taken from SunOS 4.x and Solaris 2.x UNIX systems, but apply in principle to any TCP/IP network.

Topics include:

- Using the DDLT
- Discovery: Where to look
- Name Service

• /etc/hosts, ping
• DNS: /etc/resolv.conf, in.named
• NIS: ypserv, ypbind, /etc/domainname
• NIS+DNS: /var/yp/Makefile
• Name service switch

• Ethernet interface: le0, ne0, etc.
• ARP: The ARP shotgun
• IP, the Internet protocol
• ICMP: Echo, Host Unreachable
• TCP and TELNET

• Host and network addressing
• Subnets
• The routing table, traceroute

Jim Hickstein started in UNIX systems administration at a company with 5000 hosts and 200 TCP/IP networks. He was hired as a software engineer but started doing systems administration in self-defense. He is now with Deer Run Associates.

Tutorials at-a-Glance     Tutorial Instructors