lam-helpfile - LAM help message file
DESCRIPTION
The lam-helpfile provides detailed error messages and sug
gestions for help on how to fix common problems. In many
places in LAM, when an error occurs, this help file is
consulted to display a detailed message of what the error
was and, when possible, suggestions on how to fix the
problem. It consists of much of the information from the
LAM FAQ (particularly in dealing with getting LAM up and
running)
At present, the following LAM tools use this help file (it
is expected that more will use it in future releases. If
you have suggestions for locations where more detailed
error messages would be helpful, please let us know):
hboot
lamboot
lamexec
lamhalt
lamnodes
mpicc (hcc)
mpiCC (hcp)
mpif77 (hf77)
mpirun
recon
tkill
tping
wipe
STRUCTURE AND SYNTAX
The help file is multiple blocks of help text separated by
single line delimiters. The delimiter lines are of the
format:
-*-programname:topicname-*-
Where programname is the general name of the program (or
group of programs) that this help message applies to, and
topicname is the specific topic that this message applies
two.
The special keyword ALL can be used for either the pro
gramname or the topicname in some cases; this is usually a
"wildcard" value where little specific information is
available.
Within the block of the message, lines that begin with a
"#" are treated as comments; they are not printed out.
Three special escape sequences can be used within the help
message:
ments that the help message is invoked with. The
"%N" string is replaced with the value of the Nth
argument from the argument list. The arguments are
passed from the LAM binaries themselves; they can
not be edited. The comments in the default LAM
help file explain how many arguments each message
is invoked with, and what each argument is.
%perror
Shows the result of the Unix perror(3) function.
%terror
Shows the result of the LAM terror() function,
which is essentially a wrapper around the Unix per
ror(3) function.
LOCATION OF HELP FILE
The exact location of the help file is configurable. This
allows system administrators and/or users to customize the
help file for their particular environment.
When LAM attempts to print an error message from the help
file, it looks for the help file in the following loca
tions (in order):
$LAMHELPFILE
$HOME/lam-helpfile
$HOME/lam-6.5.4-helpfile
$HOME/etc/lam-helpfile
$HOME/etc/lam-6.5.4-helpfile
$LAMHELPDIR/lam-helpfile
$LAMHELPDIR/lam-6.5.4-helpfile
$LAMHOME/etc/lam-helpfile
$LAMHOME/etc/lam-6.5.4-helpfile
$TROLLIUSHOME/etc/lam-helpfile
$TROLLIUSHOME/etc/lam-6.5.4-helpfile
$SYSCONFDIR/lam-helpfile
$SYSCONFDIR/lam-6.5.4-helpfile
Note the two variables $LAMHELPFILE and $LAMHELPDIR; these
variables can be set according to platform, for example,
to provide operating system-specific information, or
information specific to particular groups of machines,
etc.
$SYSCONFIDIR is typically $prefix/etc, where $prefix is
the location to where LAM was installed; it was the option
supplied to ./configure when LAM was built (or
/usr/local/lam-6.5.4, by default). However, note that the
value of $SYSCONFDIR can be overridden when LAM is config
ured with the --sysconfdir switch.
The following is an example customization of the help for
the hboot and lamboot programs, when the user supplies a
host file name that is not found.
-*-boot:open-hostfile-*-
%1 could not open the hostfile "%2" for the following reason:
%perror
Things to check:
- ensure that the file exists
try "ls -l %2"
- ensure that you have read permissions on the file
try "cat %2"
You may not need to specify a host file at all; the system
administrators have defined the all of Beowulf cluster host names in
the LAM default host name list. If you wish to use all of the Beowulf
nodes, simply execute:
%1 -v
If you have any problems with LAM, please send mail to:
lam-admin@your.domain.com
FILES
$LAMHOME/etc/lam-6.5.4-helpfile
default LAM help file
SEE ALSO
hboot(1), lamboot(1), lamexec(1), lamhalt(1), lamnodes(1),
mpicc(1), mpiCC(1), mpif77(1), mpirun(1), recon(1),
tkill(1), tping(1), wipe(1), perror(3)
Man(1) output converted with
man2html