conf (5)


NAME

conf - LAM node process schema

DESCRIPTION

Most LAM/MPI used can disregard this page.

A process schema lists the system processes that will constitute the LAM environment on a particular node. LAM developers will find process schemata very useful for debugging and for generating custom systems.

Separate default configurations files are normally used for booting either the single-daemon version (conf.lam) or the external-servers version (conf.otb). The single-daemon version is used by default. The external-servers version is typically used by LAM developers when debugging LAM services. The -c option to hboot(1) specifies the file name of a custom process schema. This capability is not available with lamboot(1).

Application programs can be booted with the LAM environment by including the program name in the process schema.

STRUCTURE AND SYNTAX

The process schema grammar defines three types of statements: comments, processes, and options.

Comments begin with # and terminate with a newline.

Process statements consist of a filename, command line arguments, and possibly options local to the process. The command line arguments are passed to the process when it is started. The process options are used by the booting tools before starting the process.

Filenames may include application programs and any of the system processes listed below.

bforward
bufferd helper, forwards remote messages.
bufferd
creates, kills, sweeps, and states buffers.
dli_inet
sends data on network connections.
dlo_inet
receives data from network connections.
echod
echoes messages; can be used to test nodes and links.
filed
serves file access.
flatd
provides symbolic access to node memory.
iod
handles stdio data of processes.
kenyad
controls and monitors processes.
kernel
coordinates message passing.
lamd
single-daemon version of all servers excluding ledd and mtvd.
ledd
controls LEDs.
loadd
loads executable files onto nodes.
mtvd
controls a pixel map display.
router
maintains routing tables.
traced
collects and transports trace data.

Processes are started in the order given in the process schema, and for LAM, the order is important. The kernel must be first.

Process options are placed inside braces {} following the command line arguments. Process options consist of a keyword followed by an assigned value in parentheses. Options that are not explicitly given have default settings. Options inside braces in process statements apply only to that process. Options outside process statements apply to all processes that appear later in the process schema, unless locally overridden. All the options are listed below:

inet_topo
a string of dli_inet or lamd options, global only, default null
rtr_topo
a string of router options, global only, default null
delay
waiting time in seconds after process boot, default 0

The inet_topo and rtr_topo options can be overridden from the command line of hboot(1) using the -I and -R options respectively.

EXAMPLES

Following are example process schemata. The first file is used for an OTB node in LAM.

##
## typical LAM process schemata
##

lamd $inet_topo

The second file is used by LAM developers to control each server as an independent process, typically during debugging.

##
## external-servers LAM process schemata
##

##
## The kernel is listed first.
##
kernel $delay

##
## daemons
##
router
kenyad
dli_inet $inet_topo
dlo_inet
bufferd
bforward
loadd
echod
flatd
filed
traced
iod

FILES

$LAMHOME/etc/lam-conf.lamd
default LAM node process schema, where $LAMHOME is the installation directory
$LAMHOME/etc/lam-conf.separate
example external-servers node process schema used when debugging LAM

SEE ALSO

hboot(1)