lamshrink(1) [redhat man page]

LAMSHRINK(1)							   LAM COMMANDS 						      LAMSHRINK(1)

NAME

       lamshrink - Shrink a LAM multicomputer.

SYNTAX

       lamshrink [-hv] [-w <delay>] [-u <userid>] <node> <hostname>

OPTIONS

       -h	     Print useful information on this command.

       -v	     Be verbose.

       <hostname>    Remove the LAM node on this host.

       <node>	     Remove the LAM node with this ID.

       -w <delay>    Notify processes on the doomed node and pause for <delay> seconds before proceeding.

       -u <userid>   Use this userid to access the host.

DESCRIPTION

       An existing LAM session, initiated by lamboot(1), can be shrunk to include less nodes with lamshrink.  One node is removed for each invoca-
       tion.  At a minimum, the node ID and the associated host name is given on the command line.  Once lamshrink completes, the node ID  is  in-
       valid across the remaining nodes.  If a different userid is required to access the host, it is specified with the -u option.

       Existing  application  processes on the target node can be warned of impending shutdown with the -w option.  A LAM signal (SIGFUSE) will be
       sent to these processes and lamshrink will then pause for the given number of seconds before proceeding with removing  the  node.   By  de-
       fault, SIGFUSE is ignored.  A different handler can be installed with ksignal(2).

       All  application processes on all remaining nodes are always informed of the death of a node.  This is also done with a signal (SIGSHRINK),
       which by default causes a process's runtime route cache to be flushed (to remove any cached information on the dead node).  If this  signal
       is  re-vectored	for the purpose of fault tolerance, the old handler should be called at the beginning of the new handler.  The signal does
       not, by itself, give the process information on which node has been removed.  One technique for getting this information is  to	query  the
       router for information on all relevant nodes using getroute(2).	The dead node will cause this routine to return an error.

   FAULT TOLERANCE
       If enabled with lamboot(1), LAM will watch for nodes that fail.	The procedure for removing a node that has failed is the same as lamshrink
       after the warning step.	In particular, the SIGSHRINK signal is delivered.

EXAMPLES

       lamshrink -v newhost n1
	   Remove LAM on newhost, known within LAM as node 1.  Report about important steps as they are done.

       lamshrink newhost n30 -w 10
	   Inform all processes on LAM node 30, which is running on newhost, that the node will be dead in 10 seconds.	Wait 10 seconds and remove
	   the node.  Operate silently.

SEE ALSO

       lamboot(1), tkill(1), ksignal(2), getroute(2)

LAM 6.5.8							  November, 2002						      LAMSHRINK(1)

LAMGROW(1)							   LAM COMMANDS 							LAMGROW(1)

NAME

       lamgrow - Extend a LAM multicomputer.

SYNOPSIS

       lamgrow [-hvd] [-cpu num] [-n nodeid] [-no-schedule] [-ssi key value] hostname

OPTIONS

       -cpu num      Indicate how many CPUs are available to LAM on the new node.

       -d	     Turn on debugging output.	This implies -v.

       -h	     Print useful information on this command.

       -n nodeid     Assign this ID to the new node.

       -no-schedule  Indicate that C and N expansion in mpirun and lamexec should not schedule on this node.

       -ssi key value
		     Send arguments to various SSI modules.  See the "SSI" section, below.

       -v	     Be verbose.

       hostname      Extend LAM with this host.

DESCRIPTION

       An  existing LAM universe, initiated by lamboot(1), can be enlarged to include more nodes with lamgrow.	One new node is added for each in-
       vocation.  At a minimum, the host name that will run the new node is given on the command line.	If a different userid is required  to  ac-
       cess the host, it is specified with the appropriate boot SSI options (see lamssi_boot(7)).

       The  new  node can be assigned any unused, non-negative identifier.  If no identifier is specified, the highest node identifier in the cur-
       rent LAM universe plus one is used.  Note that lamboot(1) always assigns node identifiers consecutively from 0.

       lamgrow can be run from any node in the current LAM universe.  Specifically -- it cannot be run from the intended new  host.   Two  invoca-
       tions  of lamgrow should not run concurrently, and the command attempts to detect this situation. The name of the host specified in lamgrow
       should not be the one which is already present in the user's LAM universe and the command attempts to detect this situation too.

       Resource managers will be the most common user of lamgrow.  When hosts become idle and a user has expressed a desire to	the  manager  that
       extra cycles should be exploited, the manager could invoke lamgrow and then launch the specified application process(es) on the new node.

EXAMPLES

       lamgrow -v newhost
	   Start  LAM  on  newhost  and add it to the existing LAM universe.  Choose the next available node identifier and report about important
	   steps as they are done.

       lamgrow -n 30 newhost
	   Start LAM on newhost with node ID 30 and add it to the existing LAM universe.  Operate silently.

FILES

       laminstalldir/etc/lam-conf.lamd	 default configuration file for LAM nodes, where "laminstalldir" is the directory where  LAM/MPI  was  in-
					 stalled.

BUGS

       It is not currently possible to specify a configuration file other than lam-conf.lamd on the remote node, even though this is possible with
       lamboot.

SEE ALSO

       lamboot(1), lamhalt(1), hboot(1), lamwipe(1), tkill(1), bhost(5), conf(5), lamssi_boot(7)

LAM 7.1.4							    July, 2007								LAMGROW(1)