304 lines
14 KiB
Plaintext
304 lines
14 KiB
Plaintext
# -*- text -*-
|
|
#
|
|
# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
# University Research and Technology
|
|
# Corporation. All rights reserved.
|
|
# Copyright (c) 2004-2005 The University of Tennessee and The University
|
|
# of Tennessee Research Foundation. All rights
|
|
# reserved.
|
|
# Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
# University of Stuttgart. All rights reserved.
|
|
# Copyright (c) 2004-2005 The Regents of the University of California.
|
|
# All rights reserved.
|
|
# Copyright (c) 2012 Cisco Systems, Inc. All rights reserved.
|
|
# Copyright (c) 2017-2020 Intel, Inc. All rights reserved.
|
|
# Copyright (c) 2022-2023 Nanook Consulting. All rights reserved.
|
|
# $COPYRIGHT$
|
|
#
|
|
# Additional copyrights may follow
|
|
#
|
|
# $HEADER$
|
|
#
|
|
# This is the US/English help file for palloc
|
|
#
|
|
[usage]
|
|
%s (%s) %s
|
|
|
|
Usage: %s [OPTION]...
|
|
PMIx Allocation Request tool
|
|
|
|
|
|
/***** General Options *****/
|
|
|
|
-h|--help This help message
|
|
-h|--help <arg0> Help for the specified option
|
|
-v|--verbose Enable typical debug options
|
|
-V|--version Print version and exit
|
|
--pmixmca <arg0> <arg1> Set MCA parameter value
|
|
|
|
--uri <arg0> Specify the URI of the server to which we are to connect, or
|
|
the name of the file (specified as file:filename) that contains that info
|
|
--namespace <arg0> Namespace of the daemon to which we should connect
|
|
--nspace <arg0> Synonym for "namespace"
|
|
--pid <arg0> PID of the daemon to which we should connect (int => PID or file:<file>
|
|
for file containing the PID
|
|
--system-server-first First look for a system server and connect to it if found
|
|
--system-server-only Connect only to a system-level server
|
|
--system-controller Connect to the system controller
|
|
--tmpdir <arg0> Set the root for the session directory tree
|
|
--connect-order <arg0> Specify search order for server connections - e.g., scheduler,
|
|
system controller, system-level server, or local server
|
|
--wait-to-connect <arg0> Delay specified number of seconds before trying to connect
|
|
--num-connect-retries <arg0> Max number of times to try to connect
|
|
|
|
--request-id <arg0> String identifier for this allocation request
|
|
-q|--queue <arg0> Scheduler queue this request should be passed to for processing
|
|
-N|--nodes <arg0> Number of nodes to be allocated
|
|
-i|--image <arg0> OS image to be provisioned on allocated nodes prior to initiating
|
|
execution of specified job, or turning nodes over to an interactive
|
|
session
|
|
-x|--exclude <arg0> Comma-delimited list of nodes that are to be excluded from
|
|
consideration for scheduling this allocation
|
|
--wait-all-nodes Wait for all nodes to be ready before starting execution of the
|
|
specified job.
|
|
-w|--nodelist <arg0> Comma-delimited list of ranges of specific nodes being requested
|
|
[e.g., host0[1-5],host128]. Can also pass the argument as a filename
|
|
using the "file:<path>" syntax. Ordering of names and/or duplicate
|
|
names are ignored.
|
|
--uid <arg0> Assign the resulting allocation to the specified user ID
|
|
--gid <arg0> Assign the resulting allocation to the specified group ID
|
|
-t|--time <arg0> Time limit on the assigned allocation.
|
|
--signal <arg0>[@arg1] Send all processes executing within the allocated session the specified signal
|
|
when it reaches the (optional) specified seconds of its end time.
|
|
-s|--share Allocated resources can be shared with other allocations.
|
|
--extend <arg0> Extend the specified existing session ID per the rest of the given request.
|
|
--shrink <arg0> Shrink the specified existing session ID per the rest of the given request.
|
|
--no-shell Immediately exit after allocating resources, without running a command.
|
|
--begin <arg0> Direct the scheduler to defer allocation until the specified time.
|
|
-I|--immediate <arg0> Exit if resources are not available within the time period specified.
|
|
-d|--dependency <arg0> Defer the start of this session until the specified dependencies have successfully
|
|
completed.
|
|
--do-not-wait Submit the allocation request to the scheduler, but do not wait for the allocation
|
|
to be assigned. Intended for use when submitting a job for batch execution.
|
|
|
|
The command returns the string session ID of the resulting allocation, or if executed with the "--do-not-wait" option, the
|
|
string session ID of the allocation request as assigned by the scheduler. Any command line arguments following the provided
|
|
options are taken as the job script or executable to be run within the allocation (once assigned) and its attendant
|
|
argv array. In this case, the allocation will terminate once the provide job script or executable completes.
|
|
|
|
The user is responsible for free'ing the returned string session ID.
|
|
#
|
|
[bad-option-input]
|
|
%s was given an option that expected a string argument:
|
|
|
|
option: %s
|
|
argument: %s
|
|
expected: %s
|
|
|
|
Please correct the option and try again.
|
|
#
|
|
[file-open-error]
|
|
%s was unable to open the specified file provided as an option:
|
|
|
|
option: %s
|
|
argument: %s
|
|
file: %s
|
|
|
|
Please correct the option and try again.
|
|
#
|
|
[bad-file]
|
|
%s was unable to read the necessary info from the provided file:
|
|
|
|
option: %s
|
|
argument: %s
|
|
file: %s
|
|
|
|
Please correct the option or the file and try again.
|
|
#
|
|
# CONNECTION OPTIONS
|
|
#
|
|
#
|
|
[uri]
|
|
Specify the URI of the DVM master, or the name of the file (specified as
|
|
file:filename) that contains that info
|
|
#
|
|
[num-connect-retries]
|
|
Max number of times to try to connect to the specified server (int)
|
|
#
|
|
[pid]
|
|
PID of the daemon to which we should connect (int => PID or file:<file>
|
|
for file containing the PID
|
|
#
|
|
[namespace]
|
|
Namespace of the daemon we are to connect to (char*)
|
|
#
|
|
[nspace]
|
|
Namespace of the daemon we are to connect to (char*) - synonym for "namespace"
|
|
#
|
|
[system-server-first]
|
|
First look for a system server and connect to it if found
|
|
#
|
|
[system-server-only]
|
|
Connect only to a system-level server - abort if one is not found
|
|
#
|
|
[system-controller]
|
|
Look for a system controller and connect to it if found
|
|
#
|
|
[tmpdir]
|
|
Define the root location for the session directory tree where the
|
|
rendezvous files can be found.
|
|
|
|
The rendezvous files contain connection information for a target
|
|
server and are located in the session directory tree. It may be necessary to point the
|
|
tool at the location where those files can be found if that location is other than the
|
|
expected default.
|
|
|
|
The root of the session directory defaults to the system temporary directory
|
|
as defined in the environment using (in precedence order) the envars TMPDIR, TEMP, and
|
|
finally TMP. In the absence of any of those variables, PMIx will default to the "/tmp"
|
|
location.
|
|
#
|
|
[wait-to-connect]
|
|
Delay specified number of seconds before trying to connect
|
|
#
|
|
[connect-order]
|
|
Comma-delimited list of attributes defining the order in which
|
|
connections should be attempted, from first to last. If the
|
|
final entry is not an "only" flag (e.g., PMIX_CONNECT_TO_SYSTEM),
|
|
then connection will default to the local server if no preceding
|
|
option succeeds. Thus, the following list:
|
|
PMIX_CONNECT_TO_SCHEDULER
|
|
PMIX_CONNECT_TO_SYS_CONTROLLER
|
|
PMIX_CONNECT_TO_SYSTEM
|
|
would first attempt to connect to the scheduler, then the system
|
|
controller, and then the local system-level server. If none of those
|
|
succeed, then the connection attempt will error out.
|
|
|
|
However, if the last entry were PMIX_CONNECT_SYSTEM_FIRST, then the
|
|
connection procedure would (after failing to connect to a local
|
|
system-level server) continue to include an attempt to connect
|
|
to any local server that accepted the connection request.
|
|
#
|
|
# QALLOC-SPECIFIC OPTIONS
|
|
#
|
|
[request-id]
|
|
String identifier for this allocation request. The request ID can be used for
|
|
subsequent query of request status and as a synonym for the eventual session ID
|
|
assigned by the scheduler. Note that a request ID that matches a previously
|
|
submitted request will be rejected by the scheduler - so care should be taken
|
|
to ensure that the ID provided is unique and not previously used.
|
|
#
|
|
[queue]
|
|
Scheduler queue this request should be passed to for processing. Most schedulers
|
|
have a number of queues for requests, each queue typically representing a priority
|
|
or class of request used in the scheduling process. Users should check the names
|
|
of the queues for their system using the "pquery" tool and submit allocation
|
|
requests accordingly.
|
|
#
|
|
[nodes]
|
|
The number of nodes to be allocated. In systems where nodes are being subdivided for
|
|
allocation (e.g., allocating individual cores or hwthreads), it may be necessary to
|
|
provide a more detailed description of the requested resources.
|
|
#
|
|
[image]
|
|
OS image to be provisioned on allocated nodes prior to initiating
|
|
execution of specified job, or turning nodes over to an interactive
|
|
session. Names of the available images can be obtained using the
|
|
"pquery" tool.
|
|
#
|
|
[exclude]
|
|
Comma-delimited list of nodes that are to be excluded from
|
|
consideration for scheduling this allocation
|
|
#
|
|
[wait-all-nodes]
|
|
Wait for all nodes to be ready before starting execution of the
|
|
specified job.
|
|
#
|
|
[nodelist]
|
|
Comma-delimited list of ranges of specific nodes being requested
|
|
[e.g., host0[1-5],host128]. Can also pass the argument as a filename
|
|
using the "file:<path>" syntax. Ordering of names and/or duplicate
|
|
names are ignored.
|
|
#
|
|
[uid]
|
|
Assign the resulting allocation to the specified user ID. Value
|
|
can be provided as either an integer or a username
|
|
#
|
|
[gid]
|
|
Assign the resulting allocation to the specified group ID. Value
|
|
can be provided as either an integer or a group name. Typically
|
|
used for determining file access permissions and for system
|
|
accounting.
|
|
#
|
|
[time]
|
|
Time limit on the assigned allocation. When the time limit is reached,
|
|
each process executing in the session is sent SIGTERM followed by SIGKILL.
|
|
The session is then terminated and all resources returned to the
|
|
scheduler for reuse. Time is specified in the usual time format of
|
|
months:days:hours:minutes:seconds, scanning from right to left (i.e., a
|
|
value of "2" equates to 2 seconds)
|
|
#
|
|
[signal]
|
|
When a session is within the specified seconds of its end time, send all
|
|
processes executing within it the specified signal. The <arg0> signal may
|
|
be a signal number or string name. The optional <arg1> time is taken in
|
|
seconds, defaulting to 60 seconds if not specified.
|
|
#
|
|
[share]
|
|
Allocated resources can be shared with other allocations.
|
|
#
|
|
[extend]
|
|
Extend the specified existing session ID per the rest of the given request.
|
|
For example, an existing session can have additional nodes allocated to
|
|
it, or could have its time limit increased.
|
|
#
|
|
[shrink]
|
|
Shrink the specified existing session ID per the rest of the given request.
|
|
For example, an existing session could have a certain number of nodes
|
|
removed (returning them to the scheduler for reassignment), or its
|
|
time reduced.
|
|
#
|
|
[no-shell]
|
|
Immediately exit after allocating resources, without running a command.
|
|
However, the session will still be created and will remain active and own
|
|
the allocated resources as long as it is active - i.e., it will be an
|
|
active session with an assigned session ID with no job executing within it.
|
|
You can submit commands against this resource allocation (e.g., using "prun")
|
|
if you specify the appropriate connection identification for this allocation.
|
|
Note that the session is subject to the normal constraints on sessions, including
|
|
time limits, so that eventually the session will terminate and the resources
|
|
will be freed, or you can terminate the session manually using the "pcancel"
|
|
command.
|
|
#
|
|
[begin]
|
|
Direct the scheduler to defer allocation until the specified time. Time may be
|
|
of the form HH:MM:SS to schedule the session to start at a specific time of day
|
|
(seconds are optional). If that time is already past, the next day is assumed.
|
|
You may also specify "midnight" or "noon", and you can have a time-of-day suffixed
|
|
with AM or PM for running in the morning or the evening. You can also say what
|
|
day the session should start by specifying a date of the form MMDDYY or MM/DD/YY
|
|
YYYY-MM-DD. Combine date and time using the usual format YYYY-MM-DD[THH:MM[:SS]].
|
|
You can also give times like "now + count" time-units, where the time-units can
|
|
be "seconds" (default), "minutes", "hours", days, or weeks, or you can ask that
|
|
the allocation be made "today" or "tomorrow".
|
|
#
|
|
[immediate]
|
|
Exit if resources are not available within the time period specified. If no argument
|
|
is given, resources must be available immediately for the request to succeed. By
|
|
default, --immediate is off, and the command will block until resources become
|
|
available. Since this
|
|
option's argument is optional, for proper parsing the single
|
|
letter option must be followed by an equal sign and the value - e.g., "-I=60"
|
|
for a time of 60 seconds or "-I30:00" for a time of 30 minutes.
|
|
#
|
|
[dependency]
|
|
Defer the start of this session until the specified dependencies have successfully
|
|
completed. The dependencies are provided as a comma-delimited list of session IDs.
|
|
All specified dependencies must be satisfied before the given allocation request
|
|
will be scheduled.
|
|
#
|
|
[do-not-wait]
|
|
Submit the allocation request to the scheduler, but do not wait for the allocation
|
|
to be assigned. Intended for use when submitting a job for batch execution. Users
|
|
should provide a string request ID to enable subsequent queries of request status.
|