Allied Telesis AT-8900 Series manual Starting off, Concrete piece of advice #1 Capture everything

Starting off

When going on-site to investigate a network problem, your guiding aim is to understand the root cause of the problem, which might be:

za cable plugged into the wrong socket

za misconfiguration of some piece of equipment

zan interoperability problem between two pieces of equipment

za hardware fault in some piece of equipment

za bug in the software of some piece of equipment.

Of course, it might not always be possible to find the root cause, but looking for it should be the central target of your debugging session.

However, it is also important to make sure that you are also capturing enough information (and the right sort of information) to enable someone else to effectively analyse what you have captured. This could be so that they can confirm your theory about the root cause of the problem, or, in the case that you were not able to nail down the root cause, so that they can develop their own theories about what the cause might be.

Concrete piece of advice #1: Capture everything

Right from the first moment you attach your terminal emulator to the console port of the first piece of equipment, ensure that the logging facility of the terminal emulator is turned on. This means that you will capture a lot of (seemingly) irrelevant stuff, but it also means that you will not fail to capture the important stuff.

Concrete piece of advice #2: Capture the initial state

It is important to approach the problem with an open mind, and start with a general survey of the equipment you are looking at, to give an initial picture of how things were at the start of debugging session. This might be the “network operating OK” state, or the “things going wrong” state, depending on what state the network was in when you arrived.

In this initial survey, you should capture a show debug from every relevant unit, and capture all the counters and module states that are relevant to the issue.

For example, if the network consists of switches, and the problem is related to PIM multicast routing, then you will want to capture the following commands:

show switch count show switch port=all

show switch port=all count show ip route

show ip route multi show ip igmp

2