| United States-English |
|
|
|
![]() |
Installing, Configuring and Administering the Kerberos Server V 2.0 on HP-UX 11i: HP 9000 Networking > Chapter 9 TroubleshootingTroubleshooting Kerberos |
|
When troubleshooting problems with Kerberos, you need a reference point to work from. For example, does the problem exist on the remote system or on the local system? However, the terms "local" and "remote" are limited in their description of complex communications, such as when a local system logs onto a remote system and then the remote system logs back onto the local system. At that point, which is the local system and which is the remote system? A better solution is to use the terms "client" and "server." The term "client" refers to a process that is requesting a service from another process. The term "server" refers to a process or host that performs operations requested by local or remote hosts that are running client processes. A typical network service consists of two co-operating programs. The client program runs on the requesting system. The server program runs on the system with which you want your system to communicate. The client program initiates requests to communicate. The server program accepts requests for communication. For example, the network service rlogin is the client program that requests a login to a remote HP-UX or UNIX system. When the request to log in is received on the remote host by inetd, inetd invokes the server program for rlogin (called rlogind) to handle the service request. The error messages generated by a service as seen on the client can be generated by the client or the server. Error messages from the client occur before a connection is completely established. Error messages from the server occur after a connection is completely established. System logging is handled differently by the security server. Each security server daemon, kadmind, kpropd, and kdcd writes to the system log (syslog) file. However, you can also configure the daemons to write the system logs to any file specified by you. However, principal database operations performed locally on the primary server using the Administrator are not recorded as these programs do not use syslog to audit their activities. The syslog daemon (syslogd) is configured using the /etc/syslog.conf file, which controls where your log files are located. For example, syslog can be configured to send messages to /usr/adm/messages. The security server daemons log an entry for each transaction and whether the transaction succeeded or failed. The number of transactions that are logged in your syslog file is determined by how you have configured the reporting levels. The syslog reporting levels used by the security server are: The Server logs information messages through syslog. The syslog file can grow large quickly if not maintained. The syslog file is specified in /etc/syslog.conf, which is typically /var/adm/messages. Check the size of this file to make sure it does not use an overwhelming amount of system disk space. If the /var partition grows to hundred percent utilization, then syslog will stop writing log messages and may even shut down active processes, that is, the daemons. Create a shell script to be executed daily or weekly by cron to check the syslog file size, partition utilization, or both, and detect any problems. Also, the syslog files should be archived regularly to a separate partition, drive, or server.
The following section describes various scenarios for potential problems. These debugs should help you troubleshoot and assist you in pinpointing a problem quickly. Table 9-2 Table of Errors Messages
|
||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||