llq - Query job status
llq - Query job status
Purpose
llq - Queries information about jobs in the LoadLeveler queues.
Syntax
llq [-?] [-H] [-v] [-W] [-x [-d]] [-s] [-l] [-b] [-w]
[-X {cluster_list | all} ] [-j joblist | joblist]
[-u userlist] [-h hostlist] [-c classlist]
[-R reservation_list] [-f category_list]
[-r category_list]
Flags
-?
Provides a short usage message.
-H
Provides extended help information.
-v
Displays the name of the command, release number, service level,
service level date, and lowest level of the operating system to run
this release.
-W
Specifies that the width of columns in tabular output will
be increased to fit the widest entry.
-x
Provides extended information about the selected job. If the
-x flag is used with the -r, -s, or -f flag, an error message
is generated.
CPU usage and other resource consumption information on
active jobs can only be reported using the -x flag if the
LoadLeveler administrator has enabled it by specifying A_ON
and A_DETAIL for the ACCT keyword in the LoadLeveler
configuration file.
Normally, llq connects with the central manager to obtain
job information. When you specify -x, llq connects to the
Schedd machine that received the specified job to get
extended job information. However, some statistics,
including those corresponding to System Priority and
q_sysprio, are available only from the central manager. Do
not use the -x option if you need these statistics.
When specified without -l, CPU usage for active jobs is
reported in the short format.
Note: Using both the -l and -x options without a joblist
specification can produce a very long report and
excessive network traffic.
-d
Displays the user-specified unfiltered job command file
keyword statements. Information is available only on jobs
submitted in a multicluster environment. You must specify
the -d flag in combination with the -x flag.
-s
Provides information on why a selected list of jobs remain
in the NotQueued, Idle, or Deferred state. Along with this
flag, users must specify a list of jobs. The user can also
optionally supply a list of machines to be considered when
determining why the jobs cannot run. If a list of machines
is not provided, the default is the list of machines in the
LoadLeveler cluster. For each job, llq determines why the
job remains in one of the given states instead of Running.
-l
Specifies that a long listing be generated for each job for
which status is requested.
-b
Shows Blue Gene jobs in short form. This is the Blue Gene
equivalent of the llq standard listing. Using this flag will
display the following fields:
BG The state of the job on the Blue Gene system.
Id The LoadLeveler job step ID.
LL The LoadLeveler state of the job step.
Owner The user ID of the job's owner.
Partition
The name of the Blue Gene partition assigned to the job.
PT The state of the Blue Gene partition assigned to
the job.
Size The number of Blue Gene compute nodes allocated for
the job.
Submitted
The time the job step was submitted to LoadLeveler.
-w
Provides AIX Workload Manager (WLM) CPU and real memory statistics
for jobs in the running state. This flag can be used with a joblist,
steplist, or a single stepid. All other flags except -h will result
in an error message.
When the -w flag is augmented with a single stepid, the -h
flag can be used in conjunction with -w to specify a single hostname.
This flag can only be used when ENFORCE_RESOURCE_USAGE is enabled in
the configuration file. Otherwise, an error message is produced.
The following statistics are displayed for every node the
job is running on:
* Current CPU resource consumption as a percentage of the total
resources available
* Total CPU time consumed in milliseconds
* Current real memory consumption as a percentage of the total
resources available
* The highest number of resident memory pages used
-X {cluster_list | all}
Indicates that you can specify the -X flag with either:
cluster_list Is a blank-delimited list of clusters where the
command is to run.
all Is the reserved word indicating that the command
is to run in all accessible clusters.
-j joblist |joblist
Is a blank-delimited list of job and step identifiers. When a job
identifier is specified, the command action is taken for all steps of
the job. At least one job or step identifier must be specified.
The format of a job identifier is host.jobid. The format of a step
identifier is host.jobid.stepid.
where:
* host is the name of the machine that assigned the job and step
identifiers.
* jobid is the job number assigned to the job when it was submitted.
* stepid is the job step number assigned to the job step when it was
submitted.
The job or step identifier may be specified in an abbreviated form,
jobid or jobid.stepid, when the command is invoked on the same machine
that assigned the job and step identifiers. In this case, LoadLeveler
will use the local machine's hostname to construct the full job or
step identifier.
The -j joblist flag is used to distinguish a joblist when specified
in combination with the any flag that supports a list.
If the -X flag is specified in combination with a joblist, the -j
flag must be specified. For example:
llq -X my_cluster1 my_cluster2 -j c94n13.2.1 c94n13.25.0
-u userlist
Is a blank-delimited list of users. Only job steps belonging to
users in this list are queried.
-h hostlist
Is a blank-delimited list of machines. If the -s flag is not
specified, only job steps managed by the Schedd on machines
in this list are queried. If the -s flag is specified, the
list of machines is considered when determining why a job
remains in the Idle state.
When the -h flag is used with the -w flag, only a single machine
name can be specified to obtain the WLM statistics for that machine.
-c classlist
Is a blank-delimited list of classes. Only job steps belonging to
classes in this list are queried.
-f category_list
Is a blank-delimited list of categories you want to query. Each
category you specify must be preceded by a percent sign. The
category_list cannot contain duplicate entries. This flag
allows you to create a customized version of the standard llq
listing. You cannot use this flag with the -l flag. The output
fields produced by this flag all have a fixed length. The
output is displayed in the order in which you specify the
categories. category_list can be one or more of the following:
%a Account number
%c Class
%cc Completion code
%dc Completion date
%dd Dispatch Date
%dh Hold date
%dq Queue date ("Submitted" date of "standard" llq output)
%fj Favored Job
%gl LoadLeveler group
%gu UNIX group
%h Hostname (first hostname if more than one machine is
allocated to the job step)
%id Step ID
%is Virtual image size
%jn Job name
%jt Job type
%nh Number of hosts allocated to the job step
%o Job owner
%p User priority
%R Reservation ID
%sn Step name
%st Status
%X Cluster name where the job is to be scheduled
%Xf Cluster name from where the job was sent
%Xk Cluster name the user requested
%Xs Cluster name from where the job was submitted
%Xu User name of the original submission
-r category_list
Is a blank-delimited list of formats (categories) you want to query.
Each category you specify must be preceded by a percent sign. The
category_list cannot contain duplicate entries. This flag allows you
to create a customized version of the standard llq listing. You
cannot use this flag with the -l flag. The output produced by this
flag is considered raw, in that the fields can be variable in length.
Output fields are separated by an exclamation point (!). The output
is displayed in the order in which you specify the formats.
category_list can be one or more of the formats listed under the -f
flag.
-R reservation_list
Is a blank-delimited list of reservation identifiers. Only job steps
bound to reservations in this list are queried. The format of a full
LoadLeveler reservation identifier is [host.]rid[.r].
where:
* host is the name of the machine that assigned the reservation
identifier.
* rid is the number assigned to the reservation when it was
created. An rid is required.
* r indicates that this is a reservation ID (r is optional).
The reservation identifier may be specified in an abbreviated form,
rid[.r], when the command is invoked on the same machine that
assigned the reservation identifier. In this case, LoadLeveler will
use the local machine's host name to construct the full reservation
identifier.
If a job step is not specified and if -u, -h, -c, or -R is not
specified, all jobs are queried.
If a job step is specified, you cannot specify -u, -h, -c, or -R,
except in the cases of -w and -s, for which the -h flag has
special meaning.
When -u, -h, -c, or -R are used in combination, the result is the
intersection of the job steps selected by each flag.
The -b flag can be used alone or with the -u flag and the joblist
argument. If used in conjunction with any other flag, an error
will occur.
You cannot specify -d, -x, or -w in combination with the -X flag.
Description
The llq command queries information about jobs in the LoadLeveler
queues.
Examples
1. This example generates the standard listing where the machine
mars has two jobs running and one job waiting:
Id Owner Submitted ST PRI Class Running On
---------------- ---------- ----------- -- --- -------- ----------
mars.498.0 brownap 5/20 11:31 R 100 silver mars
mars.499.0 brownap 5/20 11:31 R 50 No_Class mars
mars.501.0 brownap 5/20 11:31 I 50 silver
3 job step(s) in query, 1 waiting, 0 pending, 2 running, 0 held,
0 preempted
The standard listing includes the following fields:
Class Job class.
Id The format of a full LoadLeveler step identifier is
host.jobid.stepid. If the llq command returns
information about a job owned by a Schedd in the same
domain, then the domain of the hostname will not
appear in the output. However, when the llq command
reports information about a job owned by a Schedd in a
different domain, the fully qualified hostname is
always included. Due to space limitations, the domain
of the host may be truncated to fit in the space
allocated to the Id field. If the domain is truncated,
a dash (-) will appear at the end to indicate that
characters have been left out. To see the full job ID,
run llq with the -l flag.
Owner User ID that the job will be run under.
PRI User priority of the job step, where the values are
defined with the user_priority keyword in the job
command file or changed by the llprio command, which
is described in TWS LoadLeveler: Using and Administering.
Running On
If running, the name of the machine the job step is
running on. This is blank when the job is not running.
For a parallel job step, only the first machine is shown.
ST For more information, see the explanations of job
states in TWS LoadLeveler: Using and Administering.
Submitted
Date and time of job submission.
2. This example generates the long listing. The long listing is
generated when you specify the -x -l flags with the llq command:
llq -l -x c271f2rp01.ppd.pok.ibm.com.16.0
The long listing includes the following fields. See the "Sample
command output" topic for sample output of long listings in TWS
LoadLeveler: Using and Administering.
Account
The account number specified in the job command file.
Adapter Requirement
Reflects the settings of the network keyword in the
job command file.
For more information on the network keyword statement,
see the "Job command file keyword descriptions" under
the "Job command file reference" topic in TWS LoadLeveler:
Using and Administering.
Allocated Hosts
The machines that have been allocated for this job step.
Args Arguments that were passed to the executable.
Blocking
Reflects the settings for the blocking keyword in the job
command file.
Blue Gene Job ID
The ID of the Blue Gene job in the Blue Gene DB2(R)database.
This field is displayed for Blue Gene jobs only.
Blue Gene Status
The state of the Blue Gene job in the Blue Gene DB2
database. This field is displayed for Blue Gene jobs only.
BG Requirements
The job step requirements on Blue Gene resources as specified
at job submission. This field is displayed for Blue Gene jobs
only.
Bulk Transfer
Indicates that the value will be Yes or No depending on
whether the application requested that the communication
subsystem use bulk transfer by specifying bulkxfer=yes in
the job command file.
Checkpoint Directory
Value of the ckpt_dir keyword.
Checkpoint File
For AIX checkpointable job steps, the file name to be used for
checkpoint data.
Checkpointable
Indicates if LoadLeveler considers the job step
checkpointable (yes, no, or interval).
Ckpt Accum Time
Accumulated time, in seconds, the job step has spent
checkpointing.
Ckpt Elapse Time
Amount of time taken to perform the last successful
checkpoint.
Ckpt Execute Dir
The directory where the job step's executable will be
saved for checkpointable jobs.
Ckpt Hard Limit
Checkpoint hard limit as specified at job step submission.
Ckpt Soft Limit
Checkpoint soft limit as specified at job step submission.
Ckpt Start Time
The start time of the current checkpoint in progress. Blank
if no checkpoint running.
Class The class of the job step as specified at job submission.
class_sysprio
The class priority of the job step, where the value is defined
in the administration file.
Cluster input file
The information format is local_pathname, remote_pathname.
where:
local_pathname
Is the full pathname of the file to be copied
from the local cluster.
remote_pathname
Is the full pathname of the file that will be
copied into the remote cluster.
Cluster output file
The information format is local_pathname, remote_pathname.
where:
local_pathname
Is the full pathname of the file that will be
copied into the local cluster.
remote_pathname
Is the full pathname of the file to be copied
from the remote cluster.
Cmd The name of the executable associated with the executable
keyword (if specified) or the name of the job command file.
Comment The comment specified by the comment keyword in the job
command file.
Completion Code
The status returned by the wait3 UNIX system call.
Completion Date
Date and time job completed or exited.
Core Hard Limit
Core hard limit as specified at job submission.
Core Soft Limit
Core soft limit as specified at job submission.
Coschedule
Indicates whether the job step is required to be
coscheduled (yes or no).
Cpu Hard Limit
CPU hard limit as specified at job submission.
Cpu Soft Limit
CPU soft limit as specified at job submission.
Data Hard Limit
Data hard limit as specified at job submission.
Data Soft Limit
Data soft limit as specified at job submission.
Dependency
Job step dependencies as specified at job submission.
Dispatch Time
The time that the job was dispatched.
Env Environment variables to be set before executable
runs. Appears only when the -x option is specified.
Err The file to be used for stderr.
Error Text
The error text in the Blue Gene job record from the
Blue Gene DB2 database. This field is displayed for
Blue Gene jobs only.
Fail Ckpt Time/Date
Time and date stamp of the last failed checkpoint.
File Hard Limit
File hard limits as specified at job submission.
File Soft Limit
File soft limit as specified at job submission.
Favored Job
Indicates whether the job has been specified to have a
higher system priority than all jobs that are not
favored (yes or no).
Good Ckpt Time/Date
Time and date stamp of the last successful checkpoint.
group_sysprio
The group priority of the job step, where the value is
defined in the administration file.
high water
The highest number of resident memory pages used. Real
Memory resource only.
Hold Job Until
Job step is deferred until this date and time.
In The file to be used for stdin.
Initial Working Dir
The directory from which the job step is run. The relative
directory from which the stdio files are accessed, if
appropriate.
Job Accounting Key
The Job Accounting Key is a unique identifier for a
LoadLeveler job step. The accounting key is stored in
the AIX accounting record for each process associated
with a LoadLeveler job step. This field can be used to
correlate AIX accounting records with LoadLeveler
accounting records. The Job Accounting Key is stored
in the history file and can be displayed using the
llsummary -l command.
This keyword is not applicable on LoadLeveler for Linux
platforms.
For more information on the Job Accounting Key, see
"Gathering job accounting data" under the "Configuring
the LoadLeveler environment" topic in TWS LoadLeveler:
Using and Administering.
Job Name
The name of the job.
Job Step ID
The job step identifier.
Large Page
Indicates whether Large Page memory should be used to
run this job step. Can be Y (use Large Page memory if
available), N (No), or M (Mandatory).
LoadLeveler Group
The LoadLeveler group associated with the job step.
Machine Speed
For a serial job step, the value associated with the
speed keyword of the machine that is running this job
step. For a parallel job step, the value associated
with the speed keyword of the first machine that has
been allocated for this job step.
Max Processors
The maximum number of processors that can be used for
this job step.
McmAffinityOptions
The MCM affinity options for the job.
Min Processors
The minimum number of processors needed for this job step.
Negotiator Messages
Informational messages for the job step if it is in
the Idle or NotQueued state.
(Node) Allocated Hosts
* The machines of this Node type that have been
allocated for this job step. The format is:
hostname:task status:adapter usage, ... ,adapter usage, \
cpu usage, ... ,cpu usage + ... +,
hostname:task status:adapter usage, ... ,adapter usage, \
cpu usage, ... ,cpu usage
* The adapter usage information has the format:
adapter name(adapter window ID, network protocol, mode, \
adapter window memory)
For information on the units used to report window
memory, see the description of the "Adapter" field
in the llstatus command under the "Commands" topic
in TWS LoadLeveler: Using and Administering.
* The CPU usage information has one of the following
formats:
CPU <cpulist>
MCMnumber:CPU <cpulist>
The cpulist is a blank-delimited list of individual CPU
IDs or CPU ranges, or a combination of both CPU IDs and
CPU ranges. The CPU range is specified as the starting
CPU ID and the ending CPU ID separated by a hyphen (-).
(Node) Name
Blank value. Reserved for future use.
(Node) Node actual
Actual number of machines of this Node type that are
used in the running of this job step.
(Node) Node maximum
Maximum number of machines of this Node type that can
be used to run this job step.
(Node) Node minimum
Minimum number of machines of this Node type required
to run this job step.
(Node) Preferences
Job step preferences as specified at job submission.
(Node) Requirements
Job step requirements as specified at job submission.
(Node/Master Task) Exec Args
The arguments passed to the master task executable.
(Node/Master Task) Executable
The executable associated with the master task.
(Node/Master Task) Num Task Inst
The number of task instances of the master task.
(Node/Master Task) Task Instance
* Task instance information has the format:
hostname:task ID:adapter usage, ... ,adapter usage
* Adapter usage information has the format:
adapter name(adapter window ID, network protocol, mode, \
adapter window memory)
For information on the units used to report window memory,
see the description of the "Adapter" field in the llstatus
command under the "Commands" topic in TWS LoadLeveler: Using
and Administering.
(Node/Task) Num Task Inst
The number of task instances.
(Node/Task) Task Instance
* Task instance information has the format:
hostname:task ID:adapter usage, ... ,adapter usage, \
cpu usage
* Adapter usage information has the format:
adapter name(adapter window ID, network protocol, mode, \
adapter window memory)
For information on the units used to report window memory,
see the description of the "Adapter" field in the llstatus
command under the "Commands" topic in TWS LoadLeveler:
Using and Administering.
* The CPU usage information has one of the following formats:
CPU <cpulist>
MCMnumber:CPU <cpulist>
The cpulist is a blank-delimited list of individual CPU
IDs or CPU ranges, or a combination of both CPU IDs and
CPU ranges. The CPU range is specified as the starting
CPU ID and the ending CPU ID separated by a hyphen (-).
Node Usage
A request that a node be shared or not shared or that
a time-slice is not shared. The user specifies this
request while submitting the job.
Notifications
The notification status for the job step, where:
always
Indicates notification is sent through the mail for
all four notification categories below.
complete
Indicates notification is sent through the mail
only when the job step completes.
error
Indicates notification is sent through the mail
only when the job step terminates abnormally.
never
Indicates notification is never sent.
start
Indicates notification is sent through the mail
only when starting or restarting the job step.
Notify User
The user to be notified by mail of a job's status.
Out The file to be used for stdout.
Outbound Schedds
The list of Schedds that have acted as the outbound Schedd
for the job. The last Schedd in the list is the current
outbound Schedd. This field only displays
multicluster-specific information that was submitted or
moved to a remote cluster.
Owner The user ID that the job will be run under.
Partition ID
The ID of the Blue Gene partition allocated for the
job. This field is displayed for Blue Gene jobs only.
Partition State
The state of the Blue Gene partition allocated for the
job. This field is displayed for Blue Gene jobs only.
Port Number
The port number for InfiniBand resources used by the
running job.
Preempt Wait Count
Specifies the number of job steps that an idle job
step must preempt before it can be started.
Preemptable
Indicates whether a job step is preemptable (yes or no).
Preferences
Job step preferences as specified at job submission.
previous q_sysprio
The previous adjusted system priority of the job step.
For more information, see the discussion of how a job's
priority affects dispatching order in TWS LoadLeveler: Using
and Administering.
q_sysprio
The adjusted system priority of the job step. For more
information, see the discussion of how a job's priority
affects dispatching order in TWS LoadLeveler: Using and
Administering.
Queue Date
The date and time that LoadLeveler received the job.
Requested Cluster
The cluster the user specified at job submission.
Requested Res. ID
The reservation identifier that a job step is requested to
be bound to, but has not yet been bound to. This field will
be set when a job submitted with a request to bind has been
successfully submitted, but the bind has not yet occurred.
The bind may never occur if either the owner of the job step
is not allowed to use the reservation, or if the reservation
does not exist.
If a job command file is used to select nodes to reserve in
a make or change reservation request and the request fails,
all steps of the job, if submitted successfully, will have
MAKERES as their Requested Res. ID and the steps will be in
the NQ state.
Requirements
Job step requirements as specified at job submission.
Reservation ID
The reservation identifier that a job step is bound to. If a
job step is not bound to any reservation, this field will be
blank.
Resource
The resource being enforced by WLM. This is either CPU or
Real Memory.
Resources
Reflects the settings for the resources keyword in the
job command file.
Restart Restart status (yes or no).
Restart From Ckpt
Indicates if a job has been restarted from an existing
checkpoint (yes or no).
Restart Same Nodes
Indicates if a job step should be restarted on the
same nodes after vacate (yes or no).
Rotate Indicates whether the scheduler is free to rotate the
requested Blue Gene partition shape in order to match
an available partition (TRUE or FALSE). This field is
displayed for Blue Gene jobs only.
RSet The RSet requirement of the job.
Rss Hard Limit
RSS hard limit as specified at job step submission.
Rss Soft Limit
RSS soft limit as specified at job step submission.
Running Host
For a serial job step, the machine that is running
this job step. For a parallel job step, the first
machine that has been allocated for this job step.
Schedd History
The list of Schedds that have acted as the Schedd host
for the job. The last one in the list is the current
Schedd host. This field only displays multicluster-specific
information.
Scheduling Cluster
The cluster name where the job is to be scheduled. This
field only displays multicluster-specific information.
Sending Cluster
The cluster name that the job was sent from when moved.
This field only displays multicluster-specific information.
Shape Allocated
The allocated shape of the Blue Gene partition for the
job. This field is defined only for running-like jobs.
This field is displayed for Blue Gene jobs only.
Shape Requested
The requested shape of the Blue Gene partition for the
job, if defined, in units of base partitions. This
field is displayed for Blue Gene jobs only.
Shell The shell to be used when the job step runs.
Size Allocated
The size of the Blue Gene partition for the job in
units of compute nodes. The size allocated is not
always identical to the requested size. This field is
displayed for Blue Gene jobs only.
Size Requested
The requested size of the Blue Gene partition for the
job in units of compute nodes. The size must be
equivalent to the size of the shape, if such is
defined. This field is displayed for Blue Gene jobs
only.
SMT requested
Indicates the required Simultaneous Multi-Threading
(SMT) state, which is defined in the job command file,
if the job step requires SMT to be turned on or off.
Valid values are yes or no.
snapshot
Current CPU or Real Memory consumption as a percentage
of the total resources available.
Stack Hard Limit
Stack hard limit as specified at job submission.
Stack Soft Limit
Stack soft limit as specified at job submission.
Starter idrss/Step Starter idrss
An integral value of the amount of unshared memory in
the data segment of a process (expressed in units of
kilobytes * seconds-of-execution).
Starter inblock/Step inblock
Number of times file system performed input.
Cumulative total.
Starter isrss/Step isrss
Depending on the Operating System, this field may
contain the integral value of unshared stack size.
Starter ixrss/Step ixrss
An integral value indicating the amount of memory used
by the text segment that was also shared among other
processes (expressed in units of kilobytes *
seconds-of-execution).
Starter majflt/Step majflt
Number of page faults (I/O required). Cumulative total.
Starter maxrss/Step maxrss
Maximum resident set size utilized. Maximum value.
Starter minflt/Step minflt
Number of page faults (reclaimed). Cumulative total.
Starter msgrcv/Step msgrcv
Number of IPC messages received. Cumulative total.
Starter msgsnd/Step msgsnd
Number of IPC messages sent. Cumulative total.
Starter nivcsw/Step nivcsw
Number of involuntary context switches. Cumulative total.
Starter nsignals/Step nsignals
Number of signals delivered. Cumulative total.
Starter nswap/Step nswap
Number of times swapped out. Cumulative total.
Starter nvcsw/Step nvcsw
Number of context switches due to voluntarily giving
up processor. Cumulative total.
Starter oublock/Step oublock
Number of times file system performed output.
Cumulative total.
Starter System Time/Step System Time
CPU system time of Starter/Step processes. Cumulative
total.
Starter Total Time/Step Total Time
CPU total time of Starter/Step processes. Cumulative
total.
Starter User Time/Step User Time
CPU user time of Starter/Step processes. Cumulative
total.
Status The status (state) of the job. For more information,
see the explanation of job states in TWS LoadLeveler:
Using and Administering.
Step Adapter Memory
The total adapter pinned memory for the job step.
Step Cpu Hard Limit
Job step CPU hard limit as specified at job submission.
Step Cpu Soft Limit
Job step CPU soft limit as specified at job submission.
Step Cpus
The total ConsumableCpus for the job step.
Step Name
The name of the job step.
Step rCxt Blocks
The number of rCxt blocks for High Performance Switch
adapters.
Step Real Memory
The total ConsumableMemory for the job step.
Step Type
Type of job step:
* Serial
* General parallel
* Blue Gene
* MPICH parallel
Step Virtual Memory
The total ConsumableVirtualMemory for the job step.
Structure Version
An internal version identifier.
Submitting Cluster
The cluster name where the job was submitted from.
This field only displays multicluster-specific
information.
Submitting Host
The name of the machine to which the job is submitted.
Submitting User
The user name that the job was submitted under. This
field only displays multicluster-specific information.
System Priority
The overall system priority of the job step, where the
value is defined by the SYSPRIO expression in the
configuration file.
Task_geometry
Reflects the settings for the task_geometry keyword in
the job command file.
total Total CPU time consumed in milliseconds. CPU resource only.
Unix Group
The effective UNIX group name.
User Priority
The priority of the job step, as specified by the user
in the job command, or changed by the llprio command.
User Space Windows
The number of switch adapter windows assigned to the
job step.
user_sysprio
The user system priority of the job step, where the
value is defined in the administration file.
Virtual Image Size
The value of the image_size keyword (if specified) or
the size of the executable associated with the
executable keyword (if specified) or the size of the
job command file.
Wall Clk Hard Limit
Wall clock hard limit as specified at job submission.
Wall Clk Soft Limit
Wall clock soft limit as specified at job submission.
Wiring Allocated
Allocated type of wiring for the Blue Gene partition.
It is either TORUS or MESH. This field is displayed
for Blue Gene jobs only.
Wiring Requested
Requested type of wiring for the Blue Gene partition.
It is TORUS, MESH, or PREFER_TORUS. This field is
displayed for Blue Gene jobs only.
3. Using the abbreviated form of jobid, this example generates a
standard listing for all job steps with jobid 12 assigned by
the local machine:
llq 12
4. This example generates a standard listing for all job steps
owned by either rich or nathan and bound to reservation 6:
llq -u rich nathan -R 6
5. This example generates an extended listing for all job steps of
class batch or class highprio, managed by the Schedd daemon on
either c94n07 or c94n09:
llq -x -c batch highprio -h c94n07 c94n09
6. The following example generates a standard listing for all job
steps bound to reservation c94n04.2.r:
llq -R c94n04.2.r
You should receive a response similar to the following:
Id Owner Submitted ST PRI Class Running On
----------------- ---------- ----------- -- --- --------- -----------
c94n04.5.0 zhong 2/8 08:17 I 50 classA
1 job step(s) in query, 1 waiting, 0 pending, 0 running, 0 held,
0 preempted
7. The following example generates a customized listing for all
job steps:
llq -f %id %o %R
You should receive a response similar to the following:
Step Id Owner Reservation ID
------------------------ ----------- ------------------------
c94n04.5.0 zhong c94n04.2.r
c94n04.4.0 zhong
2 job step(s) in queue, 1 waiting, 0 pending, 0 running, 1 held,
0 preempted
8. The following is sample output for llq -X cluster2. The output
representing a cluster is delineated with a cluster header, in
this example it is cluster2.
================ Cluster cluster2 ===================================
Id Owner Submitted ST PRI Class Running On
--------------- ---------- ----------- -- --- --------- ----------
c188f2n02.9.0 brownap 10/29 13:54 R 50 april c188f2n08
1 job step(s) in query, 0 waiting, 0 pending, 1 running, 0 held,
0 preempted
9. The following is sample output for llq -X all command where there
are two clusters:
================ Cluster cluster1 ===================================
Id Owner Submitted ST PRI Class Running On
--------------- -------- ---------- -- --- --------- ----------
c193n13.283.0 rolf 8/4 10:58 R 50 No_Class c193n13
c193n13.283.0 rolf 8/4 10:58 R 50 No_Class c193n13
c193n13.289.0 rolf 8/4 10:58 R 50 No_Class c193n13
c193n13.291.0 rolf 8/4 10:59 R 50 No_Class c193n13
c193n13.293.0 rolf 8/4 10:59 R 50 No_Class c193n13
c193n13.295.0 rolf 8/4 10:59 R 50 No_Class c193n13
c193n13.297.0 rolf 8/4 11:01 R 50 No_Class c193n13
c193n13.299.0 brownap 8/4 11:02 R 50 No_Class c193n13
7 job step(s) in queue, 0 waiting, 0 pending, 7 running, 0 held,
0 preempted
================ Cluster cluster2 ===================================
Id Owner Submitted ST PRI Class Running On
--------------- -------- ---------- -- --- --------- ----------
c193n13.265.0 llbld 8/2 13:42 R 50 No_Class c197blade3b14
c193n13.267.0 llbld 8/2 13:43 R 50 No_Class c197blade3b14
c193n13.271.0 llbld 8/2 13:55 I 50 No_Class
c193n13.273.0 llbld 8/4 10:53 I 50 No_Class
c193n13.275.0 llbld 8/4 10:53 I 50 No_Class
c193n13.277.0 llbld 8/4 10:53 I 50 No_Class
c193n13.279.0 llbld 8/4 10:58 I 50 No_Class
7 job step(s) in queue, 5 waiting, 0 pending, 2 running, 0 held,
0 preempted
10. The following is sample output for llq -b for a standard Blue
Gene listing:
Id Owner Submitted LL BG PT Partition Size
-------------- ---------- ----------- --- --- --- ----------- -----
fen01.193.0 jdoe 2/21 22:52 R C part031 65536
fen02.1305.0 jfoe 2/20 02:36 R R B part031 1024
fen01.194.0 jane 2/21 14:12 I 512
11. This example generates a customized and formatted standard listing:
llq -f %id %c %dq %dd %gl %h
You should receive output similar to the following:
Step Id Class Queue Date Disp. Date LL Group Running On
------- --------- ----------- ----------- ---------- ---------------
ll6.2.0 No_Class 04/08 09:19 04/08 09:21 No_Group ll6.pok.ibm.com
ll6.1.0 No_Class 04/08 09:19 04/08 09:21 No_Group ll6.pok.ibm.com
ll6.3.0 No_Class 04/08 09:19 04/08 09:21 No_Group ll5.pok.ibm.com
3 job step(s) in queue, 0 waiting, 0 pending, 3 running, 0 held,
0 preempted
12. This example generates a customized, unformatted (raw) standard
listing. Output fields are separated by an exclamation point (!).
llq -r %id %c %dq %dd %gl %h
You should receive output similar to the following:
ll6.pok.ibm.com.2.0!No_Class!10/16/2006 09:19!10/16/2006 09:21! \
No_Group!ll6.pok.ibm.com
ll6.pok.ibm.com.1.0!No_Class!10/16/2006 09:19!10/16/2006 09:21! \
No_Group!ll6.pok.ibm.com
ll6.pok.ibm.com.3.0!No_Class!10/16/2006 09:19!10/16/2006 09:21! \
No_Group!ll5.pok.ibm.com
13. This example generates a WLM CPU and real memory statistics
listing where c209f1n05.13.0 is a CPU intensive parallel job
step currently running on the 2 nodes c209f1n05 and c209f1n01:
If the LoadLeveler interface to AIX Workload Manager (WLM) is
enabled, the -w option can be used to obtain CPU and real
memory statistics of job steps in running state. Note that
Large Page memory information is not included in the statistics
since WLM does not manage Large Page memory.
llq -w c209f1n05.13.0
You should receive output similar to the following:
=============== Job Step c209f1n05.ppd.pok.ibm.com.13.0 ============
c209f1n05.ppd.pok.ibm.com:
Resource: CPU
snapshot: 99
total: 80172
Resource: Real Memory
snapshot: 1
high water: 2561
c209f1n01.ppd.pok.ibm.com:
Resource: CPU
snapshot: 100
total: 79303
Resource: Real Memory
snapshot: 1
high water: 1919
14. The following is sample llq -l output for task instances and
allocated hosts if the job requested MCM affinity:
Allocated Hosts: e189f4rp04.ppd.pok.ibm.com:: \
sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM0:CPU< 0-5 >,MCM0:CPU< 0-5 > \
+ e189f4rp03.ppd.pok.ibm.com:: \
sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM1:CPU< 4-5 >,MCM1:CPU< 4-5 >
Num Task Inst: 4
Task Instance: e189f4rp04:0:sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM0:CPU< 0-5 >
Task Instance: e189f4rp04:1:sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM0:CPU< 0-5 >
Task Instance: e189f4rp03:2:sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM1:CPU< 4-5 >
Task Instance: e189f4rp03:3:sn1(MPI,IP,-1,Shared,0 rCxt Blks), \
MCM1:CPU< 4-5 >
Note that the < > notation will be used to list individual CPU
IDs instead of the CPU count ( ) notation when the RSET_SUPPORT
configuration file keyword is set to RSET_CONSUMABLE_CPUS or
RSET_MCM_AFFINITY.
15. The following example shows the Resource Set information in
the llq -l listing when the consumable CPUs Resource Set
requirement is requested:
Allocated Hosts : e189f4rp01.ppd.pok.ibm.com:: \
sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >,CPU< 0-5>
+ e189f4rp02.ppd.pok.ibm.com:: \
sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >,CPU< 0-5 >
Num Task Inst: 4
Task Instance: e189f4rp01:0:sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >
Task Instance: e189f4rp01:1:sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >
Task Instance: e189f4rp02:2:sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >
Task Instance: e189f4rp02:3:sn0(MPI,IP,-1,Shared,0 rCxt Blks), \
CPU< 0-5 >
16. The following example shows output for the llq -l command when
rCxt blocks are present:
.
.
.
Adapter Requirement: (sn_single,MPI,US,shared,AVERAGE,instances=1, \
rcxtblks=5)
.
.
.
Num Task Inst: 4
Task Instance: c60f1rp02:0:sn1(MPI,US,10,Shared,5 rCxt Blks),
Task Instance: c60f1rp02:1:sn0(MPI,US,10,Shared,5 rCxt Blks),
Task Instance: c60f1rp02:2:sn1(MPI,US,11,Shared,5 rCxt Blks),
Task Instance: c60f1rp02:3:sn0(MPI,US,11,Shared,5 rCxt Blks),
17. The following example shows output for the llq -l command when
SMT is requested:
=============== Job Step blablahome.clusters.com.22.0 ===============
Job Step Id: blablahome.clusters.com.22.0
Job Name: blablahome.clusters.com.22
.
.
.
Status: Running
.
.
.
Large Page: N
Coschedule: no
SMT requested: yes
Checkpointable: no
.
.
.
18. The following example shows the port number for the InfiniBand
resources used by the running job:
.
.
.
Task
----
Num Task Inst: 2
Task Instance: c171f6sq08:0:ib0(MPI,US,2,Shared,0 rCxt Blks,1), \
ib1(MPI,US,66,Shared,0 rCxt Blks,2),
Task Instance: c171f6sq08:1:ib0(MPI,US,3,Shared,0 rCxt Blks,1), \
ib1(MPI,US,67,Shared,0 rCxt Blks,2),
.
.
.
Security
LoadLeveler administrators and users can issue this command.
last change 14.02.2007 |
Michael Stephan | Print