At the top of the protocol stack are the applications and services that the user needs. Most errors are reported when a user attempts to use a service and the attempt fails. To be complete, testing needs to include the services.
Usually, a service can be tested directly. To test a web server, connect to the server with your browser. To test an FTP server, use the ftp command to connect to that server. These are user-oriented services, so they come with user-oriented commands that you can use for your testing.
Some services, however, are intended to provide service to a remote computer instead of a remote user. Directly testing these services is slightly more difficult, but it is often possible by using telnet to connect directly to the server port. You have already seen multiple examples of this in earlier chapters. Refer to the examples of testing imapd and the POP daemon in Chapter 11, "More Mail Services."
If ping tests succeed, but tests involving a specific service fail, the problem is in the configuration of the service at either the client or server end. Client configuration is usually simple and easy to check, but the server configuration can be very complex. Some complex services, such as sendmail and DNS, include their own test programs. Chapter 5, "Configuring a Mail Server," provides detailed examples of using sendmail -bt to test the local sendmail configuration. The next section of this chapter looks at the tools that are available to test a DNS configuration.
Was this article helpful?