[comp.os.vms] Managing cluster

carl@CITHEX.CALTECH.EDU (Carl J Lydick) (04/05/88)

 > In short, TO REBOOT A VAXSTATION ONE MUST BE PHYSICALLY PRESENT AT THE
 >  VAXSTATION TO TYPE IN THE SHUTDOWN/BOOT COMMANDS! Imagine trying to manage
 >  VAXstations in a large department! ( Only one locked door is sufficient to
 >  make things impossible. )
 > 
 > Perhaps I've overlooked some essential function in DECnet or perhaps someone
 >  has found a solution to this particular cluster management problem? I haven't
 >  considered the RSM ( Remote System Manager ) as that is expensive, I believe.

Try the following:
	Create a file, SYS$MANAGER:DETSHTDWN.COM containing:
		$ @SYS$SYSTEM:SHUTDOWN 0 1 0 1 0 1 R
	or substitute your own favorite shutdown parameters, but make sure
	that parameter 6 is 1.
	Then execute the DCL command:
		$ RUN/UIC=SYSTEM/INPUT=SYS$MANAGER:DETSHTDWN.COM-
			/OUTPUT=SYS$MANAGER:DETSHTDWN.LOG SYS$SYSTEM:LOGINOUT
This creates a detached process that does the system shutdown, specifying an
automatic reboot.  Since the process is detached, it won't be affected by: 
	1)  Shutting down LAT services;
	2)  Shutting down the DECnet; or
	3)  Shutting down batch queues
so it should do what you want.  The technique works for me when I have to
reboot a 780 that I can reach only by DECnet when I'm working from home. 
Please let me know if this solves your problem, and if not, why not.

								-Carl J Lydick

INTERnet:	CARL@CITHEX.CALTECH.EDU (CARL@[192.12.19.5)
BITNET:		CARL@CITHEX
HEPnet/SPAN:	CITHEX::CARL (42044::CARL)

KARCHER%MIT.MFENET@NMFECC.ARPA (04/06/88)

In response to John Macallister's query on Managing a LAVC of remote
Vaxstations: 

Managing a LAVC of remote vaxstations does have it's momements when you can't
get to a particular machine. While SHUTDOWN won't run in Batch mode it will run
from a detached job. We've used a procedure that executes on a batch queue of
the node to be rebooted and runs LOGINOUT to create a detatched job that in turn
runs SHUTDOWN. Of course if the reboot fails for some reason your sunk. The
procedure we use (35 lines) follows: 

----------------------------------------------------------------------------
$! FILE: REBOOT.COM     VERSION 1.0             EDIT: 880317 - CAK ORIG: TWF
$! DCL procedure to run shutdown to reboot LAVC nodes. It assumes batch queues
$! in the LAVC have names of format "node$BATCH". Change the SUBMIT line for
$! other conventions. The shutdown is done immediately
$
$ THIS_FILE = F$ENV("PROCEDURE")                                                ! Get the name of this file
$ NODE  := 'F$GETSYI("SCSNODE")'
$ IF F$MODE() .EQS. "BATCH" THEN GOTO BATCH                                     ! If BATCH mode dispatch to commands
$ IF F$ENV("DEPTH") .EQ. 0 THEN GOTO DETACHED                                   ! If Detached dispatch to commands
$
$! At this point we're Not Batch or Detached, assume it's interactive.
$
$ IF P1 .EQS. "" THEN INQUIRE P1 "Node to reboot [''NODE']"
$ IF P1 .EQS. "" THEN P1 = NODE
$ SUBMIT/NOPRINT/NOLOG/USER=SYSTEM/QUE='P1'$BATCH 'THIS_FILE'                   ! Submit this file on proper queue
$ EXIT
$
$! The following section runs in batch mode on the node we want to shutdown.
$! It creates a detached process that takes it's input from this the DETACHED
$! section of this file to run the actual shutdown procedure. This is required
$! since the SHUTDOWN procedure will only run in interative mode.

$BATCH:
$ LOG_FILE = F$PARSE (".LOG",THIS_FILE)                                         ! Build log filename from this file
$ RUN/DETACH/INPUT='THIS_FILE' /OUTPUT='LOG_FILE' SYS$SYSTEM:LOGINOUT           ! Create a process to run SHUTDOWN
$ EXIT
$
$! The following section supplies DCL commands to the process we created above
$! using LOGINOUT and runs the SHUTDOWN procedure on the node to be shutdown.
$DETACHED:
$ VERIFY = F$VERIFY(0)                                                          ! Turn off verification
$ WRITE SYS$OUTPUT "%REBOOT-I-STARTED, Detached job started on node ''NODE' at ''F$TIME()'"
$ REPLY/NODE='NODE'/USER/BELL "%REBOOT-I-STARTED, System shutdown procedure started ..."
$ DEFINE SHUTDOWN$INFORM_NODES 'NODE'                                           ! Only bother this node
$ @SYS$SYSTEM:SHUTDOWN 0 REBOOT NO YES SHORTLY YES REBOOT_CHECK                 ! Execute SHUTDOWN to reboot this node
------------------------------------------------------------------------------
END OF PROCEDURE

C. A. Karcher
MIT Plasma Fusion Center

KARCHER%PFCVAX@MIT-XX.LCS.MIT.EDU.ARPA
KARCHER%MIT.MFENET@NMFECC.ARPA
KARCHER%MIT.MFENET@ANLVMS.BITNET