UGCS is a medium-sized system with a lot to learn about. This guide attempts to present a logical order for learning how to administer UGCS. See UGCS Best Practices for more information.

Contents 1 Core infrastructure 1.1 AFS 1.2 LDAP 1.3 Kerberos 2 Important Services 2.1 Logging 2.2 Configurator 2.3 Cfengine

Core infrastructure

AFS

AFS is one of the most important pieces of the UGCS puzzle. You should skim the AFS user's guide. You should also bookmark the AFS administrator's guide to refer to when you have questions. If you would like temporary AFS administrator to learn about volumes, ask.

When you're done, you should have an understanding of:

1. How AFS volumes work
2. How to create/destroy AFS volumes
3. How AFS decides where to look for volumes
4. How to mount an AFS volume
5. How to create read-only copies

LDAP

LDAP is another important service that you'll need a grasp on before doing other stuff. If you learn how to use ldapsearch and ldapmodify, you should be set.

LDAP runs on hera (primary) and zeus (secondary)

When you're done, you should have an understanding of:

1. What LDAP is
2. What ldap stores
3. How to query LDAP using ldapsearch
4. How to modify LDAP entries using ldapmodify

Kerberos

Kerberos is the third piece of infrastructure that everything else relies on. Read the Moron's Guide to Kerberos You should also learn about kadmin, and change your password using it as well as learning how to look up information from it.

When you're done, you should have an understanding of:

1. How kerberos works
2. How to use kadmin

Important Services

Logging

All UGCS logs are directed to charon, our log server. Make sure you have access to https://logs.ugcs.caltech.edu, and then learn how to use splunk to query logs. You should also have an idea of what is in /var/log/ugcs on charon so you can look at the raw log files yourself.

When you're done, you should have an understanding of:

1. How to use splunk

Configurator

There are many configuration files that require a lot of the same information about each machine to be entered. Since we have a lot of machines, it would be tedious and error-prone to make them by hand. See the Configurator documentation and learn how to play around with it. You can try adding machines (remember to undo *all* changes), and look through existing scripts to see how the configuration files are made.

When you're done, you should have an understanding of:

1. How many of our configuration files are made

Cfengine