# -*- text-*-
#
# Copyright 1998-2001, University of Notre Dame.
# Authors: Jeffrey M. Squyres, Arun Rodrigues, and Brian Barrett with
# Kinis L. Meyer, M. D. McNally, and Andrew Lumsdaine
#
# This file is part of the Notre Dame LAM implementation of MPI.
#
# You should have received a copy of the License Agreement for the Notre
# Dame LAM implementation of MPI along with the software; see the file
# LICENSE. If not, contact Office of Research, University of Notre
# Dame, Notre Dame, IN 46556.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted subject to the conditions specified in the
# LICENSE file.
#
# THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
# IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
# INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
# HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
# STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
# IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
# POSSIBILITY OF SUCH DAMAGE.
#
# Additional copyrights may follow.
#
#
# ASCII help messages for LAM
#
-*-ALL:unknown-*-
# This is invoked when we have no idea what happened
# %1 = Name of the program
%1: Oops -- some unknown error happened here. Not really sure what
happened to cause this.
This might be helpful -- here's what Unix thinks the error is:
Unix errno: %errno
%perror
-*-ALL:system-call-fail-*-
# Invoked when some system call fails. This is not good!
# %1 = name of system call that failed
LAM encountered an error when invoking the system call "%1".
This is an unexpected error; we don't have much additional information
here. Perhaps this Unix error message will help:
Unix errno: %errno
%perror
-*-ALL:lib-call-fail-*-
# Invoked when some system call fails. This is not good!
# %1 = name of system call that failed
LAM encountered an error when invoking the library call "%1".
This is an unexpected error; we don't have much additional information
here. Perhaps this Unix error message will help:
Unix errno: %errno
%perror
-*-ALL:deny-root-*-
# This message is displayed when any LAM program is invoked by root.
# No parameters are sent.
It is a Very Bad Idea to run this program as root.
LAM was designed to be run by individual users; it was *not* designed
to be run as a root-level service where multiple users use the same
LAM daemons in a client-server fashion.
Especially with today's propensity for hackers to scan for root-owned
network daemons, it could be tragic to run this program as root.
While LAM is known to be quite stable, and LAM does not leave network
sockets open for random connections after the initial setup, several
factors should strike fear into system administrator's hearts if LAM
were to be constantly running for all users to utilize:
1. LAM leaves a unix domain socket open on each machine in the
/tmp directory. So if someone breaks into root on one
machine, they effectively have root on all machines that
are connected via LAM.
2. Indeed, there must have been a .rhosts (or some other trust
mechanism) for root which must have allowed you to run LAM
on remote nodes. Depending on your local setup, this may
not be safe.
3. LAM has never been checked for buffer overflows and other
malicious input types of errors. We don't *think* that
there are any buffer-overflow types of situations in LAM,
we've never checked explicitly (hence, per Mr. Murphy,
there are certainly some hiding somewhere).
4. LAM programs are not audited or tracked in any way. This
could present a sneaky way to execute binaries without log
trails (especially as root).
Hence, it's a Very Bad Idea to run LAM as root. Please login as a
different user and run LAM again.
-*-ALL:root-warning-*-
# This message is displayed when recon is invoked by root.
# No parameters are sent.
Warning: You are invoking this program as root. All other LAM executables
are not allowed to be run as root for security reasons. The use of recon
is allowed because of the common need for system administrators to validate
a LAM installation.
-*-ALL:uninitialized-*-
# This message is displayed when some program invokes an MPI function
# (not including MPI_Initialized) before invoking MPI_Init
# %1 is the hostname
# %2 is the PID
It seems that at least one rank invoked some MPI function before
invoking MPI_INIT. The only information that I can give is that it
was PID %2 on host %1.
-*-ALL:already-finalized-*-
# This message is displayed when some program invokes an MPI function
# (not including MPI_Initialized) after invoking MPI_Finalize
# %1 is the hostname
# %2 is the PID
# %3 the rank in MPI_COMM_WORLD
It seems that at least one rank invoked some MPI function after
invoking MPI_FINALIZE. The only information that I can give is that
it was PID %2 on host %1.
It was probably rank %3 on MPI_COMM_WORLD, but I can't say that for
sure...
-*-ALL:unable-rmdir-*-
# This message is displayed if there is some error while trying to remove
# a directory. This is typically in tkill when trying to remove the
# internal LAM session directory. If you see this message, something is
# very wrong, and probably requires manual intervention.
# %1 name of program with problem
# %2 directory that can't be removed
It seems that LAM was not able to remove a directory properly. This should
not happen, and will probably require manual intervention on your part.
%1 was trying to remove the following directory:
%2
when some error occurred. Perhaps the following Unix error message will
help:
Unix errno: %errno
%perror
This directory contains some internal temporary LAM files, and is safe
to remove. Please go remove it manually.
-*-lam-temp-files:badness-*-
# When we fail to create the /tmp/lam-* directory and/or files for
# temporary data in this run-time environment instance, display this
# error message.
# %1 = name of upper-level LAM library call/program that had a problem
At lamboot time, LAM stores some run-time files for each user under
the /tmp directory. This error was caused by the fact that LAM either
could not create a directory (or files in that directory), or was not
able to remove an old directory that already exists (or any files in
that previously-existing directory).
This error occurred in: %1
Things to check:
- /tmp exists and has permissions 0777
- No files/directories exist in /tmp that begin with "lam-"
and have your username in them that you do not have
permissions to remove
-*-recon:usage-*-
# Show the usage options of the "recon" program
# No parameters are sent
Synopsis: recon [-adhv] [<bhost>]
Description: Check if LAM can be booted
Options:
-a Report all errors (as opposed to quitting after the first
error)
-b Use the faster recon algorithm; will not work for users
who use the Bourne or Korn shells (bash *does* work)
-d Print debugging messages (implies -v)
-h Print this message
-v Be verbose
<bhost> Use <bhost> as the boot schema
-*-boot:bhostparse-*-
# Invoked by any booting program (e.g., lamboot, recon) when the
# user-provided host file could not be parsed properly.
# %1 = name of the invoking program
# %2 = name of the host file that we tried to parse
%1 could not parse the hostfile "%2" for the following reason:
%terror
-*-boot:close-hostfile-*-
# Invoked by any booting program (e.g., lamboot, recon) when the
# user-provided host file could not be closed properly
# %1 = name of the invoking program
# %2 = name of the host file
%1 could not close the hostfile "%2" for the following reason:
Unix errno: %errno
%perror
-*-boot:open-hostfile-*-
# Invoked by any booting program (e.g., lamboot, recon) when the
# user-provided host file could not be opened
# %1 = name of the invoking program
# %2 = name of the host file
%1 could not open the hostfile "%2" for the following reason:
Unix errno: %errno
%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"
-*-boot:resolv-hostname-*-
# Invoked by any booting program (e.g., lamboot, recon) when one of
# the hosts listed in the hostfile cannot be resolved to an IP
# address.
# %1 = name of the invoking program
# %2 = name of the host that we had a problem with
# %3 = name of the host file
%1 could not resolve the hostname "%2" that was in the host file
"%3".
Things to check:
- is "%2" in /etc/hosts?
try "grep %2 /etc/hosts"
- is "%2" resolvable by DNS (or some other naming service)?
try "ping %2" or "nslookup %2"
-*-boot:resolv-unknown-*-
# Invoked by any booting program (e.g., lamboot, recon) when some
# unknown error occurred while trying to resolve hosts in the host
# file.
# %1 = name of the invoking program
# %2 = the errno
# %3 = name of the host file
%1 encountered some unknown error (errno=%2) when trying to resolve
the hosts from the hostfile "%3". Here's a guess at the problem:
%terror
-*-boot:duplicated-host-*-
# Invoked by any booting program (e.g., lamboot, recon) when a
# duplicate host name is discovered in a host file.
# %1 = name of the invoking program
# %2 = the duplicated host from the host file
%1 found at least one duplicated host in the list (which is not
allowed):
%2
Please remove the duplicate entry for this host (and any other
duplicates) from the list.
Be sure to check for multiple names that refer to the same host (i.e.,
resolve to the same IP address).
If you are trying to execute multiple copies on an SMP, you do *not*
need to list hosts more than once. Instead, only list each host once
in the boot schema, and see the man page for mpirun(1) for the "-c"
option, which will execute multiple copies of an executable on a
single node.
-*-boot:no-localhost-*-
# Invoked when the name of the localhost was not included in the boot
# schema.
# %1 = name of the invoking program
# %2 = name of the hostfile/boot schema
%1 found that your local host is not in the hostfile "%2".
The local host name *must* be in the list of hosts in the hostfile.
In other words, you must boot LAM from a node that will be part of the
multicomputer.
- If you simply forgot to put the localhost in the boot
schema file, add it and re-run %1
- If you are trying to boot LAM from a node that will not be
part of the multicomputer, you must login to on of the nodes
that will be part of the multicomputer (i.e., one of the
nodes in the hostfiles), and re-run %1
Although the local host name is usually the first in the list to avoid
I/O ambiguities, it can actually appear anywhere in the list.
-*-boot:resolve-localhost-*-
# Invoked when there are more than one hosts in the host file and one
# resolves to 127.0.0.1. Protects users from accidently setting up an
# environment where remote lamds can not talk to the origin lamd.
# %1 = name of the invoking program
# %2 = name of offending host
%1 found that the hostname "%2" resolves to 127.0.0.1.
When using LAM/MPI with more than one host, all hosts specified in the
boot schema must resolve to an address that is reachable from all
other hosts. This is not possible with the special IP address
127.0.0.1 -- by definition, 127.0.0.1 can *only* be reached from the
local host.
In order to fix this problem:
- Ensure that "%2" is not specified in your boot
schema when attempting to %1 more than one host, or
- Ensure that the hostname
%2
does not resolve to 127.0.0.1.
Note that in many recent Linux distributions, the host name of a
machine will be added to /etc/hosts for the entry 127.0.0.1. This is
a shortcut that, while it works in many situations, is not technically
correct. The host name of the local machine should instead be in a
seperate entry with the IP address to which the name resolves.
Also note that it is perfectly acceptable to use 127.0.0.1 when only
*ONE* host is being used with LAM/MPI.
-*-boot:socket-fail-*-
# This is invoked from lambootagent.c when we failed to create an
# internet server socket
# No parameters are passed
LAM failed to generate an internet socket. Perhaps this will help:
Unix errno: %errno
%perror
-*-boot:fork-fail-*-
# This is invoked when _lam_few fails in lambootagent.c. It is only
# for the case where the local lamd failed to launch.
# %1 = name of the lamd executable that was attempted to be invoked
LAM failed to fork/exec a process to launch the local LAM daemon
(lamd). LAM first launches %1 to launch the local LAM daemon, so
several things could have gone wrong:
- "%1" itself could not be found (check your $PATH)
- "%1" failed for some reason (consult previous error messages,
if any)
- Too many processes exist and Unix could not fork another
-*-boot:remote-shell-fail-*-
# This is invoked when inetexec fails in lambootagent.c. It is only
# for the case where a remote lamd failed to launch, specifically when
# were were trying to obtain the remote shell name (i.e., before we
# tried to execute any LAM binaries)
# %1 = remote hostname/username
# %2 = local RSH attempted to be executed
# %3 = remote command attempted to be executed
# %4 = sample command to execute
LAM failed to execute a process on the remote node "%1".
LAM was not trying to invoke any LAM-specific commands yet -- we were
simply trying to determine what shell was being used on the remote
host.
LAM tried to use the remote agent command "%2"
to invoke "%3" on the remote node.
This usually indicates an authentication problem with the remote
agent, or some other configuration type of error in your .cshrc or
.profile file. The following is a list of items that you may wish to
check on the remote node:
- You have an account and can login to the remote machine
- Incorrect permissions on your home directory (should
probably be 0755)
- Incorrect permissions on your $HOME/.rhosts file (if you are
using rsh -- they should probably be 0644)
- You have an entry in the remote $HOME/.rhosts file (if you
are using rsh) for the machine and username that you are
running from
- Your .cshrc/.profile must not print anything out to the
standard error
- Your .cshrc/.profile should set a correct TERM type
- Your .cshrc/.profile should set the SHELL environment
variable to your default shell
Try invoking the following command at the unix command line:
%4
You will need to configure your local setup such that you will *not*
be prompted for a password to invoke this command on the remote node.
No output should be printed from the remote node before the output of
the command is displayed.
When you can get this command to execute successfully by hand, LAM
will probably be able to function properly.
-*-boot:remote-stderr-*-
# This is invoked when inetexec fails in lambootagent.c by getting
# something printed to the stderr from a remote shell.
# %1 = remote hostname/username
# %2 = local RSH attempted to be executed
# %3 = remote command attempted to be executed
# %4 = sample command to execute
LAM attempted to execute a process on the remote node "%1",
but received some output on the standard error.
LAM tried to use the remote agent command "%2"
to invoke "%3" on the remote node.
This can indicate an authentication error with the remote agent, or
can indicate an error in your $HOME/.cshrc, $HOME/.login, or
$HOME/.profile files. The following is a list of items that you may
wish to check on the remote node:
- You have an account and can login to the remote machine
- Incorrect permissions on your home directory (should
probably be 0755)
- Incorrect permissions on your $HOME/.rhosts file (if you are
using rsh -- they should probably be 0644)
- You have an entry in the remote $HOME/.rhosts file (if you
are using rsh) for the machine and username that you are
running from
- Your .cshrc/.profile must not print anything out to the
standard error
- Your .cshrc/.profile should set a correct TERM type
- Your .cshrc/.profile should set the SHELL environment
variable to your default shell
Try invoking the following command at the unix command line:
%4
You will need to configure your local setup such that you will *not*
be prompted for a password to invoke this command on the remote node.
No output should be printed from the remote node before the output of
the command is displayed.
When you can get this command to execute successfully by hand, LAM
will probably be able to function properly.
-*-boot:no-shell-*-
# This is invoked when we fail to get a shell output from the remote node in
# inetexec.c. That is, the length of the output is zero.
# %1 = remote hostname/username
# %2 = local RSH attempted to be executed
# %3 = remote command attempted to be executed
# %4 = sample command to execute
LAM was trying to determine the your shell on the "%1".
However, LAM did not receive any valid output.
LAM tried to use the remote agent command "%2"
to invoke "%3" on the remote node.
This is an unusaual error -- it does not typically indicate a
permissions problem. But it can sometimes indicate latent (or
"silent") errors in your $HOME/.cshrc, $HOME/.login, or $HOME/.profile
file.
Try invoking the following command at the unix command line:
%4
You will need to configure your local setup such that you will *not*
be prompted for a password to invoke this command on the remote node.
No output should be printed from the remote node before the output of
the command is displayed.
When you can get this command to execute successfully by hand, LAM
will probably be able to function properly.
-*-boot:remote-boot-fail-*-
# This is invoked when inetexec fails in lambootagent.c. It is only
# for the case where a remote lamd failed to launch, specifically when
# we tried to execute some LAM binary.
# %1 = remote hostname/username
# %2 = local RSH attempted to be executed
# %3 = remote shell
# %4 = sample command to execute
LAM failed to execute a LAM binary on the remote node "%1".
Since LAM was already able to determine your remote shell as "%3",
it is probable that this is not an authentication problem.
LAM tried to use the remote agent command "%2"
to invoke the following command:
%4
This can indicate several things. You should check the following:
- The LAM binaries are in your $PATH
- You can run the LAM binaries
- The $PATH variable is set properly before your
.cshrc/.profile exits
Try to invoke the command listed above manually at a Unix prompt.
You will need to configure your local setup such that you will *not*
be prompted for a password to invoke this command on the remote node.
No output should be printed from the remote node before the output of
the command is displayed.
When you can get this command to execute successfully by hand, LAM
will probably be able to function properly.
-*-bhost-parse:garbage-after-hostname-*-
# Called when something after a hostname is not a key-value pair
# %1 = Line number
# %2 = Line contents
An invalid entry was found after the hostname on hostfile line %1.
This indicates either: there were two hostnames on one line, or a
key-value pair did not have valid syntax.
Only one hostname is allowed per line, and key-value pairs must be of
the form "key=value", with no spaces. The line will be ignored.
The offending line was:
%2
-*-bhost-parse:hostname-must-be-first-*-
# Called when a key-value pair precedes a hostname
# %1 = Line number
# %2 = Line contents
A key-value pair was found where a hostname was expected on line %1 of
the hostfile.
The first word on each hostfile line must be a hostname. The line
will be ignored.
The offending line was:
%2
-*-bhost-parse:invalid-cpu-count-*-
# Called when a CPU count is not a number, or is <=0
# %1 = Line number
# %2 = Line contents
In line %1 of the hostfile, an invalid CPU count was found. The CPU
count must be a valid positive integer. Also, there must be no spaces
between the equals sign and the number. The line will be ignored.
The offending line was:
%2
-*-bhost-parse:multiple-cpu-counts-*-
# Called when more than one CPU count was found
# %1 = Line number
# %2 = Line contents
In line %1 of the hostfile, multiple CPU counts were found. Only one
CPU count is allowed per line of the hostfile. The line will be
ignored.
The offending line was:
%2
-*-bhost-parse:empty-username-*-
# Called when a user name is empty
# %1 = Line number
# %2 = Line contents
In line %1 of the hostfile, an empty user name was found. There must
be no spaces between the equals sign and the user name. The line will
be ignored.
The offending line was:
%2
-*-bhost-parse:multiple-usernames-*-
# Called when more than one username was found
# %1 = Line number
# %2 = Line contents
In line %1 of the hostfile, multiple usernames were found. Only one
user name is allowed per line of the hostfile. The line will be
ignored.
The offending line was:
%2
-*-bhost-parse:invalid-key-*-
# Called when an invalid key is found
# %1 = Line number
# %2 = Line contents
In line %1 of the hostfile, an invalid key was found. Currently, the
only keys supported are "cpu" for the number of CPUs, and "user" for
the username to use for logging in to the host. The line will be
ignored.
The offending line was:
%2
-*-recon:happiness-*-
# Invoked when recon completes successfully.
# No parameters are sent.
Woo hoo!
recon has completed successfully. This means that you will most likely
be able to boot LAM successfully with the "lamboot" command (but this
is not a guarantee). See the lamboot(1) manual page for more
information on the lamboot command.
If you have problems booting LAM (with lamboot) even though recon
worked successfully, enable the "-d" option to lamboot to examine each
step of lamboot and see what fails. Most situations where recon
succeeds and lamboot fails have to do with the hboot(1) command (that
lamboot invokes on each host in the hostfile).
-*-recon:unhappiness-*-
# Invoked when recon completes unsuccessfully.
# No parameters are sent.
recon was not able to complete successfully. There can be any number
of problems that did not allow recon to work properly. You should use
the "-d" option to recon to get more information about each step that
recon attempts.
Any error message above may present a more detailed description of the
actual problem.
Here is general a list of prerequesites that *must* be fullfilled
before recon can work:
- Each machine in the hostfile must be reachable and operational.
- You must have an account on each machine.
- You must be able to rsh(1) to the machine (permissions
are typically set in the user's $HOME/.rhosts file).
*** Sidenote: If you compiled LAM to use a remote shell program
other than rsh (with the --with-rsh option to ./configure;
e.g., ssh), or if you set the LAMRSH environment variable
to an alternate remote shell program, you need to ensure
that you can execute programs on remote nodes with no
password. For example:
unix% ssh -x pinky uptime
3:09am up 211 day(s), 23:49, 2 users, load average: 0.01, 0.08, 0.10
- The LAM executables must be locatable on each machine, using
the shell's search path and possibly the LAMHOME environment
variable.
- The shell's start-up script must not print anything on standard
error. You can take advantage of the fact that rsh(1) will
start the shell non-interactively. The start-up script (such
as .profile or .cshrc) can exit early in this case, before
executing many commands relevant only to interactive sessions
and likely to generate output.
-*-lamboot:usage-*-
# Show the usage options of the "lamboot" program
# No parameters are sent
Synopsis: lamboot [-dhHvVx] [<bhost>]
Description: Start up a LAM multicomputer
Options:
-b Use the faster boot algorithm; will not work for users
who use the Bourne or Korn shells (bash *does* work)
-d Print debugging messages (implies -v)
-h Print this message
-H Don't print the header
-s Close stdout/stderr of the locally-launched LAM daemon
-v Be verbose
-V Print version and exit without booting
-x Run nodes in fault tolerant mode
<bhost> Use <bhost> as the boot schema
-*-wipe:usage-*-
# Show the usage options of the "wipe" program
# No parameters are sent
Synopsis: wipe [-dhHvV] [-n <#>] [<bhost>]
Description: This command has been obsoleted by the "lamhalt" command.
You should be using that instead. However, "wipe" can
still be used to shut down a LAM multicomputer.
Options:
-b Use the faster wipe algorithm; will not work for users
who use the Bourne or Korn shells (bash *does* work)
-d Print debugging message (implies -v)
-h Print this message
-H Don't print the header
-n <#> Wipe the first <#> nodes
-v Be verbose
-V Print version and exit without shutting down LAM
<bhost> Use <bhost> as the boot schema
-*-lamboot:about-to-wipe-*-
# Invoked when lamboot failed and is about to do a wipe to kill
# all nodes that it already successfully booted.
# No parameters are sent.
lamboot encountered some error (see above) during the boot process,
and will now attempt to kill all nodes that it was previously able to
boot (if any).
Please wait for LAM to finish; if you interrupt this process, you may
have LAM daemons still running on remote nodes.
-*-lamboot:wipe-fail-*-
# Invoked in lamboot.c when the call to _lam_few() fails when we are
# trying to tkill
# No parameters are sent.
lamboot tried to execute the "wipe" program on the local node to
remove any already-booted LAM nodes, but failed.
Things to check:
- Is "wipe" in your $PATH?
- Did "wipe" fail for some reason (consult previous error messages,
if any)
- Did Unix fail to fork the "wipe" program due to lack of
resources?
-*-hboot:usage-*-
# Show the usage options of the "hboot" program
# No parameters are sent
Synopsis: hboot [-dhnNstv] [-c <schema>] [-I <inet_topo>] [-R <rtr_topo>]
Description: Start LAM on the local node
Options:
-c <conf> Use <conf> as the process schema
-b <name> Use <name> for the unix socket names
-d Print debug information (implies -v)
-h Print this message
-I <inet_topo> Set $inet_topo variable
-N Pretend to hboot
-R <rtr_topo> Set $rtr_topo variable
-s Close stdio of processes
-t Kill existing session first
-v Be verbose
-*-hboot:psc-parse-*-
# Invoked from hboot.c when psc_parse returns an error
# %1 = name of boot schema file that did not parse
hboot could not parse the hostfile "%1" for the following reason:
Unix errno: %errno
%perror
-*-hboot:nothing-to-do-*-
# Invoked from hboot.c when there is nothing to do
# %1 = name of boot schema file
hboot could not find anything to do in the hostfile "%1".
Things to check:
- Is the file empty?
- Does the file contain no valid hostnames?
-*-hboot:cant-find-executables-*-
# Invoked from hboot.c when we can't find an executable
# %1 = name of executable that we couldn't find
hboot could not find the executable "%1" in your path.
Things to check:
- Is "%1" in your $PATH?
- Does your .cshrc/.profile set your $PATH properly?
The following Unix error message may be helpful:
Unix errno: %errno
%perror
-*-hboot:tkill-fail-*-
# Invoked from hboot.c when tkill failed to run on the local node
# Called with no parameters
hboot was not able to run "tkill" on the local node.
Things to check:
- Is "tkill" in your $PATH?
- Does your .cshrc/.profile set your $PATH properly?
The following Unix error message may be helpful:
Unix errno: %errno
%perror
-*-tkill:usage-*-
# Show the usage options of the "tkill" program
# No parameters are sent
Synopsis: tkill [-dhNv] [-f <killfile>]
Description: Terminate LAM on the local node. This command is usually
invoked only internally in LAM.
Options:
-d Print debug information.
-b <name> Use <name> for the unix socket names
-h Print this message.
-N Pretend to tkill.
-f <kill> Use <kill> as the kill file.
-v Be verbose.
-*-tkill:cannot-kill-*-
# Invoked when tkill fails to kill a process that it thinks is still
# there
# %1 = PID of process that its trying to kill
tkill failed to kill the LAM daemon; I think that it is PID %1, but
"kill" and "kill -9" did not seem to kill it.
Things to check:
- Do a "ps" and see if the process still exists
- Use the Unix kill(1) command to kill the process
-*-compile:phome-too-long-*-
# Invoked from hcc.c/hf77.c when the LAM directory is too long
# %1 = what LAMHOME (or TROLLIUSHOME or DEFP) is
# %2 = max num of characters that it can be
# %3 = which compiler was invoked
The path name given to %3 for use with the -I and -L switches was too
long. It can be a maximum of %2 characters long. The pathname given was:
%1
The pathname was given either in the $LAMHOME environment variable,
$TROLLIUSHOME environment variable, or as the prefix when LAM was
configured.
-*-compile:no-profiling-support-*-
# Invoked from hcc.c/hf77.c when -lpmpi appears on the command line,
# but LAM was not configured/compiled with profiling support.
# %1 = name of compiler invoked
%1: "-lpmpi" was included on the command line indicating use of MPI profiling
%1: layer. However, LAM/MPI was not compiled with support for the MPI
%1: profiling layer; please contact your system administrator to re-install
%1: LAM/MPI.
-*-hf77:no-fortran-support-*-
# Invoked when hf77 was invoked by the user, but LAM was not compiled
# with Fortran support.
Sorry, LAM was not compiled with FORTRAN support.
You will need to re-compile LAM and be sure that either ./configure
finds a FORTRAN compiler, or that you explicitly specify a FORTRAN
compiler with the --with-fc switch to ./configure.
-*-mpirun:no-init-*-
# This is invoked when mpirun received a no-init message back from one
# of the programs that it started did not seem to call MPI_Init
# %1 = Node of the process that did not invoke MPI_Init
It seems that [at least] one of processes that was started with mpirun
did not invoke MPI_INIT before quitting (it is possible that more than
one process did not invoke MPI_INIT -- mpirun was only notified of the
first one, which was on node n%1).
mpirun can *only* be used with MPI programs (i.e., programs that
invoke MPI_INIT and MPI_FINALIZE). You can use the "lamexec" program
to run non-MPI programs over the lambooted nodes.
-*-mpirun:abnormal-exit-*-
# Invoked when one of the processes started by mpirun fails
# %1 pid of the failed process
# %2 node that the failed process was on
# %3 exit status of the failed process
One of the processes started by mpirun has exited with a nonzero exit
code. This typically indicates that the process finished in error.
If your process did not finish in error, be sure to include a "return
0" or "exit(0)" in your C code before exiting the application.
PID %1 failed on node n%2 with exit status %3.
-*-mpirun:no-impi-support-*-
# Invoked when someone gives -client or -server to mpirun, but
# LAM was not configured/compiled with IMPI support.
You selected "-client" or "-server" on the command line, indicating
that you'd like to run an IMPI job, but LAM/MPI was not compiled with
IMPI support. See your administrator if you want LAM/MPI installed
with IMPI support.
This flag has been ignored, and mpirun will continue.
-*-mpirun:no-impi-server-*-
# Invoked when someone gives -server option to mpirun, but LAM was not
# configured with support for the IMPI server.
You selected "-server" on the command line, indicating that you'd like
to run the IMPI server, but LAM/MPI was not compiled support for the
IMPI server. See your administrator if you want LAM/MPI installed
with support for the IMPI server.
This flag has been ignored, and mpirun will continue.
-*-mpirun:server-without-client-*-
# Invoked when a user specifies the -server mpirun option without also
# specifying the -client option.
You have used the "-server" option to mpirun without also using the
"-client" option. LAM currently does not support running the server
without running a client as well. If you wish to run the server by
itself (i.e., without a LAM client), you may invoke it independantly
of LAM.
-*-mpirun:cant-find-impi-server-*-
# Invoked when the user specifies -server to mpirun, but mpirun is not
# able to find the "impi-server" binary.
The "-server" option was specified to mpirun, but mpirun was not able
to locate the "impi-server" binary in your path. Check your path for
the location of the impi-server binary.
-*-mpirun:usage-*-
# Usage message for mpirun. No arguments.
Synopsis: mpirun [options] <app>
mpirun [options] <where> <program> [<prog args>]
Description: Start an MPI application in LAM/MPI.
Notes:
[options] Zero or more of the options listed below.
<app> LAM/MPI appschema.
<where> List of LAM nodes and/or CPUs (examples
below).
<program> Must be a LAM/MPI program that either
invokes MPI_INIT or has exactly one of
its children invoke MPI_INIT.
<prog args> Optional list of command line arguments
to <program>.
Options:
-c <num> Run <num> copies of <program> (same as -np).
-c2c Use fast library (C2C) mode.
-client <rank> <host>:<port>
Run IMPI job; connect to the IMPI server <host>
at port <port> as IMPI client number <rank>.
-D Change current working directory of new
processes to the directory where the
executable resides.
-f Do not open stdio descriptors.
-ger Turn on GER mode.
-h Print this help message.
-lamd Use LAM daemon (LAMD) mode (opposite of -c2c).
-nger Turn off GER mode.
-np <num> Run <num> copies of <program> (same as -c)
-nsigs Don't catch signals.
-nx Don't export LAM_MPI_* environment variables.
-O Multicomputer is homogeneous.
-pty Use pty if stdout is a tty.
-s <nodeid> Load <program> from node <nodeid>.
-toff Enable tracing with generation initially off.
-ton, -t Enable tracing with generation initially on.
-v Be verbose.
-w / -nw Wait/don't wait for application to complete.
-wd <dir> Change current working directory of new
processes to <dir>.
-x <envlist> Export environment vars in <envlist>.
Nodes: n<list>, e.g., n0-3,5
CPUS: c<list>, e.g., c0-3,5
Extras: h (local node), o (origin node), N (all nodes), C (all CPUs)
Examples: mpirun n0-7 prog1
Executes "prog1" on nodes 0 through 7.
mpirun -lamd -x FOO=bar,DISPLAY N prog2
Executes "prog2" on all nodes using the LAMD RPI.
In the environment of each process, set FOO to the value
"bar", and set DISPLAY to the current value.
mpirun n0 N prog3
Run "prog3" on node 0, *and* all nodes. This executes *2*
copies on n0.
mpirun C prog4 arg1 arg2
Run "prog4" on each available CPU with command line
arguments of "arg1" and "arg2". If each node has a
CPU count of 1, the "C" is equivalent to "N". If at
least one node has a CPU count greater than 1, LAM
will run neighboring ranks of MPI_COMM_WORLD on that
node. For example, if node 0 has a CPU count of 4 and
node 1 has a CPU count of 2, "prog4" will have
MPI_COMM_WORLD ranks 0 through 3 on n0, and ranks 4
and 5 on n1.
mpirun c0 C prog5
Simlar to the "prog3" example above, this runs "prog5"
on CPU 0 *and* on each available CPU. This executes
*2* copies on the node where CPU 0 is (i.e., n0).
This is probably not a useful use of the "C" notation;
it is only shown here for an example.
Defaults: -c2c -w -nger -pty
-*-lamexec:usage-*-
# Usage message for lamexec. No arguments.
Synopsis: lamexec [options] <app>
lamexec [options] <where> <program> [<prog args>]
Description: Start a non-MPI job on LAM nodes.
Notes:
[options] Zero or more of the options listed below.
<app> LAM/MPI appschema.
<where> List of LAM nodes and/or CPUs (examples
below).
<program> Program to execute.
<prog args> Optional list of command line arguments
to <program>.
Options:
-c <num> Run <num> copies of <program> (same as -np).
-D Change current working directory of new
processes to the directory where the executable
resides.
-f Do not open stdio descriptors.
-h Print this help message.
-pty Use pty if stdout is a tty.
-np <num> Run <num> copies of <program> (same as -c).
-npty Do not use a pty for stdout.
-s <nodeid> Load <program> from node <nodeid>.
-v Be verbose.
-w / -nw Wait / don't wait for job to complete.
-x <envlist> Export environment vars in <envlist>.
Nodes: n<list>, e.g., n0-3,5
CPUS: c<list>, e.g., c0-3,5
Extras: h (local node), o (origin node), N (all nodes), C (all CPUs)
Examples: lamexec n0-7 prog1
Executes "prog1" on nodes 0 through 7.
lamexec -x FOO=bar,DISPLAY N prog2
Executes "prog2" on all nodes. In the environment of
each process, set FOO to the value "bar", and set
DISPLAY to the current value.
lamexec n0 N prog3
Run "prog3" on node 0, *and* all nodes. This executes *2*
copies on n0.
lamexec C prog4 arg1 arg2
Run "prog4" on each available CPU with command line
arguments of "arg1" and "arg2". If each node has a
CPU count of 1, the "C" is equivalent to "N". If at
least one node has a CPU count greater than 1, LAM
will run multiple copies on that node. For example,
if node 0 has a CPU count of 4 and node 1 has a CPU
count of 2, "prog4" will have MPI_COMM_WORLD ranks 0
through 3 on n0, and ranks 4 and 5 on n1.
lamexec c0 C prog5
Simlar to the "prog3" example above, this runs "prog5"
on CPU 0 *and* on each available CPU. This executes
*2* copies on the node where CPU 0 is (i.e., n0).
This is probably not a useful use of the "C" notation;
it is only shown here for an example.
Defaults: -w -pty
-*-lamnodes:usage-*-
# Usage message for lamnodes. No arguments.
Synopsis: lamnodes [options] [<where> ...]
Description: Print host names of LAM nodes. If no nodes are
specified, print host names of all nodes in the LAM.
Uses gethostbyaddr(3) to get remote host names. If
gethostbyaddr(3) fails prints, the IP address of the node.
Options: -c Suppress printing CPU count for each node
-h Print this message.
-i Print IP addresses (instead of IP names)
-n Suppress printing LAM node numbers
Examples: lamnodes
Print names of all nodes in the LAM. This is
equivalent to "lamnodes N".
lamnodes -n h
Print name of local node, with no LAM node number.
lamnodes -i c0 n2 -c
Print IP numbers of the node where CPU 0
resides, and node 2. Do not print the CPU
count for each node.
lamnodes C -c -n
Print the IP names of the nodes where each
CPU resides, and suppress printing the CPU
count. This will produce a valid boot schema,
suitable for use with lamboot.
lamnodes N -n
Print the IP names of all nodes along with
their CPU count. Like the above example, this
can be used to produce a boot schema that can
be used with lamboot.
-*-lamhalt:usage-*-
# Usage message for lamhalt. No arguments.
Synopsis: lamhalt [options]
Description: Shut down an entire running LAM run time system.
This command should be preferred over the "wipe" command;
it is much faster and does not require a boot schema
(hostfile). Instead, it simply tells the local LAM
daemon to shutdown and waits for acknowledgements.
Options: -h Print this message.
-H Don't print the header
-v Be verbose
-d Print out lots of debugging messages
-*-tping:usage-*-
# Usage message for tping. No arguments.
Synopsis: tping [-vh] [-c <count>] [-d <delay>] [-l <length>] <where>
Description: Echo messages to Trollius nodes. Useful for testing
if LAM daemons are working properly.
Options: -h Print this help message.
-v Turn off verbose mode.
-c <#> Echo <#> messages.
-l <#> Use <#> bytes per message.
-d <#> Delay <#> seconds between messages.
Nodes: n<list>, eg., n0-3,5,0xa,12-15
CPUs: c<list>, eg., c0-3,5,0xa,12-15
h (local node), o (origin node), N (all nodes), C (all CPUs)
Example: tping n10 -c 100 -l 1024 -d 0
Echo 100 messages of length 1024 to node 10
without pausing between messages.
Notes: Specifying more than one CPU on a single node is not
useful; LAM will translate this into a single reference
that node.
-*-impid:usage-*-
Synopsis: impid client_rank num_hosts server_host server_port
or impid client_rank num_hosts server_host:server_port
Description: Start the LAM IMPI client daemon
Options: client_rank integer (0-31) specifying rank on IMPI server
num_hosts integer specifying local MPI_COMM_WORLD size
server_host IP name or number specifying location of server
server_host port number where server is listening
The IMPI_AUTH_KEY environment variable must be set to a valid numeric
key, or the IMPI_AUTH_NONE environment variable must exist (but this
is highly discouraged).
***********************************************************************
*** This program should probably only be run automatically from LAM ***
***********************************************************************
-*-impi-client:possible-loopback-*-
# Invoked when the impid detects 127.x.y.z from its localhostname
# %1: address in question
# %2: hostname that we got %1 from
WARNING: The LAM IMPI client has detected the address %1 from the
hostname %2.
It is likely that this is a loopback address, and will not work as
expected. LAM will try this address anyway, but if the IMPI jobs
hangs during the setup phase, check your DNS and hostname settings.
-*-impi-client:different-private-nets-*-
# Invoked when the local IMPI host is 192.168.something or
# 10.something, and a remove IMPI host is 10.something or
# 192.168.something, respectively (i.e., they're potentially on
# different private networks)
# %1: local IP address
# %2: local hostname
# %3: remote IP address
# %4: function name (accept or connect)
WARNING: The LAM IMPI client (hostname %2, IP address %1)
is trying to %4() to IP address %3.
These try addresses appear to be on different private networks. LAM
will try anyway, but it is possible that the operation will eventually
time out and fail.
If these addresses are not correct, check the DNS and hostname
settings on the node with the LAM IMPI client, as well as the nodes
that the other IMPI hosts are running on (each IMPI host reports its
own IP address; if they determine their IP address incorrectly, other
IMPI hosts may have difficulty contacting them). Addtionally, check
to ensure that both hosts are reachable (in both directions) from each
other.
-*-impi-client:local-private-net-*-
# Displayed when the IMPI host detects that it is a 192.168.x.y or
# 10.x.y.z, but one of the other hosts's IP addresses is not.
# %1: local IP address
# %2: local IP hostname
# %3: remote IP address
# %4: function (accept or connect)
WARNING: The LAM IMPI host (on %2) has determined that it is on
a private network (%1), and that at least one other IMPI host is
on a public network (%3).
LAM will still try to connect these two hosts, but it is likely that
the connection will eventually timeout and fail. If it does, check
the DNS and hostname settings on %2 and %3 to see if they are
reachable (in both directions) from each other.
-*-impi-client:remote-private-net-*-
# Displayed when the IMPI host detects that one or more of its host
# addresses is a 192.168.x.y or 10.x.y.z, its local address is not.
# %1: local IP address
# %2: local IP hostname
# %3: remote IP address
# %4: function (accept or connect)
WARNING: The LAM IMPI host (on %2) has determined that it is on
a public network (%1), and that at least one other IMPI host is
on a private network (%3).
LAM will still try to connect these two hosts, but it is likely that
the connection will eventually timeout and fail. If it does, check
the DNS and hostname settings on %2 and %3 to see if they are
reachable (in both directions) from each other.
-*-impi-client:no-auth-*-
# Message invoked when no authentication mechanism can be found by
# the IMPI client.
WARNING: No IMPI authentication method can be found! It is highly
likely that the IMPI server will reject this connection.
The available IMPI authentication methods are:
IMPI_AUTH_KEY - Set the "IMPI_AUTH_KEY" environment variable to a numeric
value. The IMPI server must have the same value.
IMPI_AUTH_NONE - Set the "IMPI_AUTH_NONE" environment variable. The use
of this mechanism is *strongly* discouraged.
The server must share one of the available authentication methods for
IMPI authentication to suceed.
-*-impi-client:unexpected-connection-*-
# Displayed when someone made a connection during the IMPI dance comes
# from an unexpected IP address.
# %1: IP address of unexpected source
# %2: IP name of unexpected source
# %3: local hostname
WARNING: The LAM IMPI host on %3 accepted a connection from an
unexpected IP address (%1 / %2).
The LAM IMPI host will now abort this MPI job because this could signal
an attack on your system.
-*-ALL:no-lamd-*-
# Message declaring that there is no lamd on the current node.
# %1 = name of command that the user tried to invoke
It seems that there is no lamd running on this host, which indicates
that the LAM/MPI runtime environment is not operating. The LAM/MPI
runtime environment is necessary for the "%1" command.
Please run the "lamboot" command the start the LAM/MPI runtime
environment. See the LAM/MPI documentation for how to invoke
"lamboot" across multiple machines.
-*-ALL:lamd-full-*-
# Message declaring that the lamd on the current host has filled up
# its descriptor table and cannot accept any more processes.
# %1 = name of program that the user tried to invoke
# %2 = max number of descriptors in lamd
# %3 = host name where lamd had problems
The lamd that is on the host "%3" is full.
That is, the maximum number of %2 local client processes have already
attached to that lamd. It will refuse new connections until one or
more of the %2 current processes die.
-*-MPI:no-lamd-*-
# Message declaring that there is no lamd on the current node and some
# MPI program was attempted (i.e., not a LAM/MPI support executable)
# %1 = name of function that the user tried to invoke
It seems that there is no lamd running on this host, which indicates
that the LAM/MPI runtime environment is not operating. The LAM/MPI
runtime environment is necessary for MPI programs to run (the MPI
program tired to invoke the "%1" function).
Please run the "lamboot" command the start the LAM/MPI runtime
environment. See the LAM/MPI documentation for how to invoke
"lamboot" across multiple machines.
-*-MPI:lamd-full-*-
# Message declaring that the lamd on the current host has filled up
# its descriptor table and cannot accept any more processes.
# %1 = name of function that the user tried to invoke (probably always
# MPI_Init)
# %2 = max number of descriptors in lamd
# %3 = host name where lamd had problems
The lamd that is on the host "%3" is full.
That is, the maximum number of %2 local client processes have already
attached to that lamd. It will refuse new connections until one or
more of the %2 current processes die.
Hence, this MPI program will now abort, and you will get an error
message from mpirun about how some program did not complete MPI_Init.
-*-MPI:bad-mpi-init-*-
# Called from MPI_Init when lam_linit fails for some non-LAM-specific
# reason.
It seems that some error has occurred during MPI_INIT. This will
cause your process to abort. These kinds of errors are usually
system-related, such as running out of disk space, running out of
memory, or something more serious such as data not being passed
between processes properly. That is, you should not be seeing this
error message; if you are, somethings is likely Very Wrong with your
system. :-(
Perhaps this Unix error message will help:
Unix errno: %errno
%perror
-*-rpi.tcp:open-server-socket-*-
# Called in the TCP RPI when we can't open a server socket to listen
# for other ranks to connect to us on.
# %1 = rank number in MPI_COMM_WORLD
It seems that rank %1 was not able to open a TCP server socket for
some reason. LAM is likely to abort your program shortly. :-(
Perhaps this unix error message will help:
Unix errno: %errno
%perror
-*-rpi.tcp:open-client-socket-*-
# Called in the TCP RPI when we can't open a client socket to some other
# rank during the connection dance.
# %1 = rank number in MPI_COMM_WORLD
It seems that rank %1 was not able to open a TCP client socket for
some reason. LAM is likely to abort your program shortly. :-(
Perhaps this unix error message will help:
Unix errno: %errno
%perror
-*-rpi.tcp:accept-server-socket-*-
# Called in the TCP RPI when accepting on a server socket during the
# connection dance fails.
# %1 = rank number in MPI_COMM_WORLD
It seems that rank %1 was not able to accept on an already open TCP
server socket for some reason. LAM is likely to abort your program
shortly. :-(
Perhaps this unix error message will help:
Unix errno: %errno
%perror
-*-rpi.myri:gm-dma-malloc-fail-*-
# Called in the myrinet RPI when it cannot allocate enough dynamic memory
# %1 = failing rank
# %2 = how much it tried to allocate
It seems that rank %1 was not able to allocate additional
DMA-accessible memory for Myrinet. DMA-accessible memory is memory
which the Myrinet cards can access. Typically, OS's have fixed limits
on how the total amount of DMA memory can be allocated at one
time. Long MPI messages require that a large amount of DMA-accessible
memory be allocated. If possible, try using smaller messages or
adjusting the OS's DMA limit. :-(
GM failed to allocate a DMA block of %2 bytes.
