This page lists a series of checks to make for a regular LTSP/Ocean/NAS/router environment. Ideally it should be a flow chart diagram, but it would be difficult to update and rather big.
Power on client machine, watch for startup messages.
YES: ➔ DHCP request
NO: Check power and video cables
➔ Computer hardware failure or electric failure.
After starting up the BIOS in the network interface card (NIC) should start a basic PXE process by requesting an IP address via DHCP. Watch the status of this request.
YES: ➔ TFTP request
NO: TIMEOUT No DHCP server found ➔ Ocean up
NO: OUT OF LEASES Too many client machines on network ☹.
NO: NO BOOT FILE Incorrect DHCP server, check for rogue DHCP servers and ➔ Ocean up
NO: Check machine case for lights on network cable
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Ocean up
After acquiring an IP address the PXE BIOS loads the Linux kernel and an initial boot image.
YES: ➔ NFS mount
NO: CRC ERROR A network error has occured, reboot and try again.
NO: TIMEOUT ➔ Ocean up
NO: Test tftpboot on Ubuntu server
$ cd /tmp $ tftp ocean tftp> get pxelinux.0 Received 13300 bytes in 0.0 seconds
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Ocean up
After loading the kernel and performing system initialisation the root file system is mounted over NFS, if this fails an error message is displayed.
GOOD: ➔ LDM login with Kerberos
BAD: ➔ NAS up
Can you login with a LTSP client using an account stored in Kerberos.
YES: ✔ Everything is working. Problems envolving system stability are not covered in this article.
NO: ➔ LDM login with passwd
Can you login with an account stored in /etc/passwd, these are accounts local to the LTSP server and should only be used for testing.
YES: Something is not correct with the Kerberos authentication, reproduce to on command line ➔ SSH login with Kerberos
NO: Something might not be correct with the LTSP configuration, confirm if regular login is OK ➔ SSH login with passwd
Login on a command line via SSH to the LTSP server. It is recommended to have a local account configured in /opt/ltsp/i386/etc/passwd to allow this to be done on a VT on the LTSP client: Ctrl+Alt+F1.
# ssh user@ubuntu
YES: Regular login works fine, further analysis is required to diagnose why LDM login fails
➔ Login to server via SSH with passwd, check /var/log/auth.log and /var/log/messages for errors from the LDM login.
➔ If no trace of login is found check /etc/lts.conf on the LTSP client to make sure it is pointing to the correct server address.
➔ Try rebooting the LTSP client and re-login via LDM.
☹ Report problem: LDM fails but SSH succeeds.
NO: ➔ SSH login with passwd
Login on a command via SSH using a local account to the LTSP server.
# ssh test@ubuntu
YES: Further analysis required on Kerberos authentication ➔ Request Kerberos ticket with kinit
NO: A more fundamental problem might be occuring, check the basics ➔ Ubuntu server up
Kerberos authentication relies on the basic function of requesting a Kerberos ticket from the Kerberos server. This function can be manually performed on the command line and any error inspected. On the Ubuntu server run the following.
# kinit user
YES: Local ticket request succeeds, now re-test SSH login
➔ SSH login to localhost with Kerberos user.
# ssh user@localhost
➔ Reconfirm SSH from LTSP client machine.
➔ Reboot LTSP client machine and re-try.
☹ Report problem: Local SSH/Kerberos succeeds but remote fails.
NO: clock skew Ubuntu and directory server times are mismatched ➔ Ocean time
NO: password expired directory server experienced a big time jump, reboot the directory server and try again.
NO: cannot find KDC The directory server cannot be contacted ➔ Ocean up
NO: incorrect password View end of /var/log/auth.log for real reason.
☹ Report problem: Kinit auth.log error.
NO:
☹ Report problem: Kinit other error.
Test if the NAS file server is up and functioning on the network:
$ ping nas
YES: Login via web interface: http://nas/
➔ Check disk and system status.
NO: Test ping from another machine.
➔ Check network light on NAS appliance.
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Check power light on NAS appliance
➔ Reboot nas and try again.
➔ Complete hardware failure: replacement required.
Test if the Ubuntu server can respond to a network ping:
# ping ubuntu
YES: Possible crash or hardware failure, reboot Ubuntu server and try again.
NO: Test ping from another machine.
➔ Check networking
➔ Ubuntu server network light
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Attach monitor and view startup messages
➔ Complete hardware failure: replacement required.
Test if the directory server can respond to a network ping:
# ping ocean
YES: Try login via web interface.
➔ Check process list, either services might have crashed or re-configured due to error.
➔ Reboot directory server and try again.
NO: Test ping from another machine.
➔ Check networking
➔ Ocean server network light
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Ocean server power light
➔ Ocean crash or hardware failure, reboot directory server and try again.
➔ Attach serial cable and monitor start up messages through a terminal program.
➔ Complete hardware failure: replacement required.
Find the current time of the directory server via the web interface: http://ocean/
CORRECT: Re-try kinit.
➔ Serious network or hardware failure, or recent software upgrade severly breaking basic functionality.
INCORRECT: Test connectivity to reference time source
$ ping pool.ntp.org
➔ Internet up ➔ Reboot directory server and try again.
Test that internet connectivity is functional.
$ ping www.news.com
YES: Failure with ISP DNS service or very rare NTP server error, wait for problem to be resolved.
DNS OK, NO PING: Check other host
$ ping 4.2.2.2
NO PING: Either ISP or router problem, further analysis required ➔ Router up
Test that the local router is functional.
$ ping router
YES: Try login into routers web interface
➔ View internet connection status in web interface
➔ Wait for resolution or contact ISP.
NO: Test ping from another machine.
➔ Test local networking
➔ Router network lights on
➔ Switch crash or hardware failure, reboot switch and try again.
➔ Router power on
➔ Router crash or hardware failure, reboot router and try again.
➔ Complete hardware failure: replacement required.
© 2008 Novell, Inc. All Rights Reserved.
