Rebooting Sun Hosts

Emergency Sun Host Reboot/Power Off

If no one with root password access (carlos, caleb, jules or arlin) is available and you have determined (see below) that a reboot is necessary, you can reboot as follows:

If host has a front panel 'button'

(ie. all SunBlades and all Ultras except for Ultra-1's) then to reboot safely:

Push and let go quickly the lit button on the front panel. The screen will go black. About a minute or two later, the screen will light up, a message about syncing the disks will be displayed and then the cpu box will be powered off.

After a moment, check that the light in or near the front panel button is off. Then push and let go quickly the button. The host will boot back up.

  • NOTE: Host lakme in Room 1325 has a hardware problem and this

does not work on that host.

  • NOTE: If the 'safe' reboot doesn't work, the last resort is

to hold in the 'button' about 4 seconds or more until the cpu box powers off (light is out). If that fails, then unplug the cpu box.

  • See below for errors that occur when the 'safe' reboot isn't possible.

If host has no front panel 'button'

(ie. Ultra-1: attila, isolde, lucia. SPARCtation-10: odyssey. SPARCstation-5: doncarlo) Then to reboot safely:

   Hold down 'Stop' key and the 'a' key.  This will display the okay prompt:
		OK>
   In order to sync the disks (this does NOT work on other hosts),
	type:
		OK> sync
   After many messages (ie: syncing disks, dumping, ...) the screen
   will go black.  When the screen turns white, the host has started
   to reboot.

   If you need to power off the host, then instead, as soon as the
   screen turns white for the reboot, do a 2nd 'Stop-A'.  Then at the
   prompt, type
		OK> power-off
   to power off the cpu box.

	NOTE: If the 'safe' reboot doesn't work, the last resort is 
	      use the switch by the plug to power off the cpu box.

   *See below for errors that occur when the 'safe' reboot isn't possible.*

Is a Reboot Necessary?

  • Most Sun hosts act as data disk servers, some also as home directory servers and others as mail, print, web, and software servers.
  • As the Sun systems are interdependent, they are ideally always up and running with all disks mounted. Rebooting should be a last resort.
  • Below are guidelines on how to avoid a reboot/poweroff of a Sun host. See guidelines for doing a 'safe' reboot/poweroff above.

How to avoid a reboot

  • Keep a 'console' window open. All system error messages are displayed in the console window. (In CDE's menu, select HOSTS --> Terminal Console )
  • Check the messages in the console window. If they include:

... server SOME_HOST not responding ...

then your host is likely just waiting on another host that is temporarily unavailable. Wait for the remote host to be back up. You should only reboot as a last resort and only if you can work without that remote host.

  • If the console messages include:
    	... server SOME_HOST_X not responding ...
    	... server SOME_HOST_Y not responding ...
    	... server SOME_HOST_Z not responding ...
       then there is likely a network problem and rebooting will not help.
       Get in touch with the system manager.
    
  • If your console session is hung or frozen and there are no

... server SOME_HOST not responding ...

messages, then try to kill the session from another host:

		> ssh -l your_username  your_hostname
		> psallw | grep ksh
USER PID  %CPU PRI NI S %MEM  VSZ   SZ  ELAPSED    TIME TT      COMMAND   
YOU  9561  0.0  59 20 S  0.1 1952  244  1-02:32:54 0:00 ?  /bin/ksh \
							/usr/dt/bin/Xsession
		> kill  9561		(PID# for your Xsession 'ksh' process)

Likely Errors After 'Unsafe' Reboot/Power Off

These occur when the 'safe' reboot/poweroff described above was not possible.

	Disk .../dev/dsk/c0t0d0s0... not OKAY

		...Run fsck...

   Before the indicated disk(s) can be mounted, the root account must run
   'fsck' on that disk.  If no one is available to use the root
   account, the disk and, possibly the host, will remain off line.