wlmpard(1M)wlmpard(1M)NAMEwlmpard - activate the HP-UX Workload Manager Global Arbiter
SYNOPSIS
[-p] [-s] [-t] [-n] [-l par[=n]] -A
[-p] [-s] [-t] [-n] [-l par[=n]]
[-n]
DESCRIPTION
Use to:
· Migrate CPU resources (cores) across virtual partitions (a core is
the actual data-processing engine within a processor, where a pro‐
cessor might have multiple cores)
· Simulate core migration between nPartitions using Instant Capacity
(formerly known as iCOD)
· Optimize use of Temporary Instant Capacity (v6 or later) and Pay
per use (v4, v7, or later) resources to minimize costs
· Manage CPU allocation for a combination of nPartitions, virtual
partitions, PSET workload groups, and FSS workload groups
Every global arbiter interval (120 seconds by default), the WLM global
arbiter checks for CPU requests.
When managing partitions, the arbiter moves cores across partitions, if
needed, to better achieve the SLOs specified in the WLM configuration
files that are active in the partitions. (Given the physical nature of
nPartitions, WLM only simulates core movement--as described in the wlm‐
parconf(4) manpage.) The WLM daemon must be running in each partition.
Also, the WLM configuration in each partition must use the keyword to
reference the name of the system where the global arbiter is running.
(This system can be one of the partitions or another system. The
global arbiter can run on any HP-UX system that has network connectiv‐
ity to the partitions being managed by WLM.)
WLM allows you to manage a combination of nPars containing vPars con‐
taining FSS or PSET-based groups.
In addition to partition management, optimizes Temporary Instant Capac‐
ity (v6 or later) and Pay per use (v4, v7, or later) resources.
A global arbiter configuration file is required.
logs messages to the file /var/opt/wlm/msglog in the partition in which
it is running.
To start automatically at system boot, edit the file /etc/rc.con‐
fig.d/wlm.
OPTIONS
Displays usage information and exits.
This option overrides all other options.
Displays version information and exits. This option overrides all
options other than
Displays the most recent global arbiter configuration, appending two
commented lines that indicate the origin of the configuration.
Prevents the global arbiter from running in daemon mode (that is,
forces it
to run in the foreground).
Causes the global arbiter to run in passive mode. In this mode, you can
see
how WLM will approximately respond to a particular global arbiter
configuration--without the configuration actually taking control
of your system. Using this mode allows you to verify various items
in your global arbiter configuration.
You can run in passive mode with each partition's daemon running
in regular mode. Thus, you can run experiments on a production
system without consequence.
To see how WLM approximately responds to the configuration, use
the WLM utility
For more information on passive mode, including its limitations,
see the wlm(5) manpage.
Causes the global arbiter to run in secure mode if you have
distributed security certificates to the systems in question. For
more information on using security certificates, see the wlm‐
cert(1M) manpage.
The global arbiter runs in secure mode by default when you use the
/sbin/init.d/wlm script to start WLM. If you upgraded WLM, secure
mode might not be the default. Ensure that the variable in the
file /etc/rc.config.d/wlm is enabled. You can change the default
by editing this variable.
Generates comma-separated audit data files. These files are placed in
the
directory /var/opt/wlm/audit/ and are named wlmpard.monyyyy, with
monyyyy representing the month and year the data was gathered. You
can access these files directly or through the command. For infor‐
mation on or on the format of the data files, see the wlmaudit(1M)
manpage.
Be sure to set in your WLM global arbiter configuration file as
indicated in the wlmparconf(4) manpage when you use the option.
The interval for the global arbiter daemon should be larger than
the largest WLM interval being used on the system.
Activates a copy of the most recent global arbiter configuration.
WLM activation may take longer than usual when managing nParti‐
tions.
Activates the configuration specified in the file
configfile. If configfile is not valid, an error message is dis‐
played, and exits.
WLM activation may take longer than usual when managing nParti‐
tions.
Checks the configuration specified in the file
configfile for syntax errors. The current configuration is not
affected.
-l par[=n]
Causes to log either virtual partition or nPartition statistics to
the file /var/opt/wlm/wlmpardstats every global arbiter interval
or every n intervals.
You can use the command to review statistics from
/var/opt/wlm/wlmpardstats:
% wlminfo par -o
For more information on see the wlminfo(1M) manpage.
You can enable automatic trimming of the wlmpardstats file by
using the keyword in your WLM global arbiter configuration. For
more information, see the wlmparconf(4) manpage.
Kills any running instance of
Use this option to shutdown the HP-UX Workload Manager global
arbiter.
WLM shutdown may take longer than usual when managing nPartitions.
HOW TO USE wlmpard TO MIGRATE CORES ACROSS PARTITIONS
HP recommends running WLM global arbitration in secure mode; otherwise,
a rogue user could manipulate the communications, resulting in one or
more partitions being granted an incorrect number of CPU resources
(cores). To enable secure communications, you must set up security cer‐
tificates and distribute them to all systems in question. For more
information, refer to the section HOW TO SECURE COMMUNICATIONS in the
wlmcert(1M) manpage.
Assuming you have completed the required steps for setting up and dis‐
tributing security certificates, WLM global arbitration runs in secure
mode by default when you use the /sbin/init.d/wlm script to start WLM.
(If you upgraded WLM, secure mode might not be the default. Ensure that
the variable in /etc/rc.config.d/wlm is enabled.) You can also acti‐
vate global arbitration in secure mode by using the option. HP recom‐
mends using secure communications. If you must disable secure communi‐
cations, use global arbitration only on trusted local area networks.
For information on disabling secure communications, refer to the HP-UX
Workload Manager User's Guide.
Do not use cell-specific core bindings or user-assigned core bindings
on virtual partitions you are going to manage with WLM.
With Instant Capacity v6 or earlier, do not include spaces in partition
names. Also, if or truncate the name of an nPartition, use to shorten
the name so that it is not truncated.
Do not adjust any partition WLM is managing while is running. This
includes using or to change the name, configuration, or resources (CPU
and memory) associated with the virtual partition or nPartition. This
also includes using to perform online cell (cell OL*) operations on a
cell in a WLM-managed partition. (To check the status of online cell
operations, use the command.) To adjust a partition or cell, first shut
down WLM (including on all partitions that will be affected by the mod‐
ification, then modify the partition. Restart WLM after modifying the
partition. (Changes to Instant Capacity affect the entire complex;
changes to a virtual partition affect the nPartition only, unless
Instant Capacity is configured on the nPartition.) For example, if WLM
is managing two virtual partitions vParA and vParB, and you need to
migrate memory resources from vParA to vParB, you must shut down WLM in
both virtual partitions. As another example, to change the name of an
nPartition, you must first shut down WLM in every operating system
instance across the entire complex, because the name change affects
Instant Capacity, and Instant Capacity changes affect every nPartition
across the complex.
You can configure WLM to manage partitions and workload groups (either
FSS or PSET-based groups) at the same time. For restrictions pertain‐
ing to HP Integrity Virtual Machines, see the section "Compatibility
with HP Integrity Virtual Machines".
The following steps explain how to use WLM's global arbiter:
1. (Optional) Set up secure WLM communications
Follow the procedure HOW TO SECURE COMMUNICATIONS in the wlmcert(1M)
manpage--skipping the step about starting/restarting the WLM dae‐
mons. You will do that later in this procedure.
2. Create a WLM configuration file for each partition
Each partition on the system must have the WLM daemon running. Cre‐
ate a WLM configuration file for each partition, ensuring each con‐
figuration uses the keyword to reference the system where the global
arbiter is running.
For example WLM configurations for virtual partitions and nParti‐
tions, see /opt/wlm/examples/wlmconf/par_usage_goal.wlm and
/opt/wlm/examples/wlmconf/par_manual_allocation.wlm.
WLM allocates cores to a partition based on the CPU resource limits
of the partition (physical limits for nPartitions; logical limits
for virtual partitions). For example, WLM adjusts the number of CPU
resources (cores) assigned to a virtual partition within the limits
of the given virtual partition's minimum and maximum number of
cores, which you set using
The way WLM uses group weighting to determine CPU allocations across
partitions is equivalent to the way it uses group weighting to
determine allocations within a partition. For more information,
refer to the wlmconf(4) manpage and the HP-UX Workload Manager's
User's Guide (/opt/wlm/share/doc/WLMug.pdf).
If the tunable is set to 0 in a WLM configuration that contains
goal-based SLOs, those SLOs may not release CPU resources properly
when the CPU resources are no longer needed. is set to 1 by
default. For more information, see the wlmconf(4) manpage.
3. Activate each partition's WLM configuration file in passive mode
if desired
WLM operates in "passive mode" when you include the option in your
command to activate a configuration. With passive mode, you can see
approximately how WLM will respond to a particular configura‐
tion--without the configuration actually taking control of your sys‐
tem.
Activate each partition's WLM configuration file configfile in pas‐
sive mode as follows:
configfile
WLM activation may take longer than usual when managing nPartitions.
To see how WLM will approximately respond to the configuration, use
the WLM utility
4. Activate each partition's WLM configuration file
After verifying each partition's WLM configuration file configfile
in passive mode, activate it as follows:
configfile
To use secure communications, activate the file using the option:
configfile
The daemon runs in secure mode by default when you use the
/sbin/init.d/wlm script to start WLM. (If you upgraded WLM, secure
mode might not be the default. Ensure that the variable in
/etc/rc.config.d/wlm is enabled.)
5. Create a configuration file for the global arbiter
On the system running the global arbiter, create a configuration
file for the global arbiter. (If this system is being managed by
WLM, it will have both a WLM configuration file and a WLM global
arbiter configuration file. You can set up and run the global
arbiter configuration on a system that is not managed by WLM if
needed for the creation of a fault-tolerant environment or a Ser‐
viceguard environment.)
This global arbiter configuration file is required. In this file,
you can specify the:
· Global arbiter interval
· Port used for communications between the partitions
· Size at which to truncate wlmpardstats log file
· Lowest priority at which Temporary Instant Capacity (v6 or
later) or Pay per use (v4, v7, or later) resources are used
Specifying this priority ensures WLM maintains compliance with
your usage rights for Temporary Instant Capacity. When your
prepaid amount of temporary capacity expires, WLM no longer
attempts to use the temporary resources.
If you specify an interval, it should be larger than the largest WLM
interval you use on the system. For the syntax of this file--as well
as default values for the items above, see the wlmparconf(4) man‐
page.
For example WLM global arbiter configurations for managing virtual
partitions and nPartitions, see /opt/wlm/examples/wlm‐
conf/par_usage_goal.wlmpar and /opt/wlm/examples/wlmconf/par_man‐
ual_allocation.wlmpar.
6. Activate the global arbiter in passive mode if desired
Similar to WLM's passive mode, the WLM global arbiter has a passive
mode (also enabled with the option) that allows you to verify set‐
tings before you let it control your system.
Activate the global arbiter configuration file configfile in passive
mode as follows:
configfile
WLM global arbiter activation may take longer than usual when manag‐
ing nPartitions.
Again, to see how WLM responds, approximately, use the WLM utility
7. Activate the global arbiter
Activate the global arbiter configuration file as follows:
configfile
To use secure communications, activate the file using the option:
configfile
The global arbiter runs in secure mode by default when you use the
/sbin/init.d/wlm script to start WLM. (If you upgraded WLM, secure
mode might not be the default. Ensure that the variable in
/etc/rc.config.d/wlm is enabled.)
In a complex (a system divided into either nPartitions or virtual
partitions):
· If the complex is an Instant Capacity system, you must use
exactly one process to manage all the partitions in the com‐
plex that are under WLM control
· If the complex is not an Instant Capacity system, you must use
a separate process for each nPartition--with each managing
only the virtual partitions within its nPartition.
A complex is an Instant Capacity system if Instant Capacity is
installed and Instant Capacity usage rights are applied on the sys‐
tem.
MANAGEMENT OF NESTED nPARTITIONS / VIRTUAL PARTITIONS /
WORKLOAD GROUPS
You can manage any combination of FSS or PSET-based workload groups
inside virtual partitions inside nPartitions if desired.
Certain software restrictions apply to using PSET-based groups with HP-
UX Virtual Partitions (vPars), Instant Capacity, and Pay per use. For
more information, refer to the WLM Release Notes
(/opt/wlm/share/doc/Rel_Notes).
For information on setting up this type of management, see the chapter
in the WLM user's guide (/opt/wlm/share/doc/WLMug.pdf) titled "Manage‐
ment of nested nPars / vPars / workload groups."
HOW TO USE wlmpard TO OPTIMIZE TEMPORARY INSTANT CAPACITYAND PAY PER USE SYSTEMS
If you have Temporary Instant Capacity (v6 or later) or Pay per use
(v4, v7, or later) resources available on a system, WLM can help you
optimize the use of those resources--using them only as needed to meet
the service-level objectives for your workloads.
While has always managed migration of CPU resources across partitions
for WLM, it also now provides management of Temporary Instant Capacity
and Pay per use resources for WLM. This management is available on
standalone systems, as well as across a collection of partitions. With
the keyword mentioned below, optimizes the Temporary Instant Capacity
or Pay per use resources. It determines the amount of resources needed
to meet your workloads' SLOs, then keeps the total number of active
cores on the system to the minimum needed to meet that resource demand.
By minimizing the number of active cores, WLM reduces your costs.
To use to optimize the use of these resources, follow the steps given
above in the section "HOW TO USE wlmpard TO MIGRATE CORES ACROSS PARTI‐
TIONS" --with one exception: Specify the keyword, which works with Tem‐
porary Instant Capacity v6 or later, and with PPU v4, v7, or later. For
more information on this keyword, see the wlmparconf(4) manpage.
RETURN VALUE
returns exit status if no errors occur, or if there are errors.
returns exit status if no errors occur or nonzero if the configuration
file is invalid or there were errors while activating the configura‐
tion.
returns exit status if no errors occur or nonzero if the running could
not be confirmed as killed.
EXAMPLES
Check the configuration file for syntax errors:
configfile
Activate the specified configuration file:
configfile
COMPATIBILITY NOTES
Compatibility with Instant Capacity
Use WLM's virtual partition management in combination with Instant
Capacity only when using vPars version A.03.01 or later. (Instant
Capacity was formerly known as iCOD.)
Use WLM's nPartition management with the Instant Capacity versions
specified in the WLM Release Notes (/opt/wlm/share/doc/Rel_Notes).
Compatibility with Temporary Instant Capacity
WLM manages any present Temporary Instant Capacity (TiCAP) whenever is
running. Use the keyword in your configuration file for to ensure WLM
maintains compliance with your usage rights for Temporary Instant
Capacity. By default, WLM will not use Temporary Instant Capacity when
15 or fewer processing days of temporary capacity are available; in
this case, you must purchase extra capacity. Before adding the capac‐
ity, be sure to stop (using the option).
You can change the 15-day default by setting the WLM global arbiter
keyword. For more information on this keyword and on the keyword, see
the wlmparconf(4) manpage.
Compatibility with PSETs
As of WLM A.03.01, you can configure WLM to manage PSET-based workload
groups and partitions (virtual partitions or nPartitions) at the same
time. Certain software restrictions apply to using PSET-based groups
with HP-UX Virtual Partitions (vPars), Instant Capacity, and Pay per
use. For more information, refer to the WLM Release Notes
(/opt/wlm/share/doc/Rel_Notes).
Compatibility with HP Integrity Virtual Machines
WLM can run on an Integrity Virtual Machines (Integrity VM) Host or
within an Integrity VM (guest). You can run WLM both on the Integrity
VM Host and in an Integrity VM, but each WLM runs as an independent
instance. To run WLM on the VM Host, you must use strictly host-based
configurations--WLM configurations designed exclusively for moving
cores across nPartitions or for activating Temporary Instant Capacity
or Pay per use cores. (WLM will not run with FSS groups or PSETs on
systems where Integrity VM is running.) In addition, ensure that the
minimum number of cores allocated to a WLM host is greater than or
equal to the maximum number of virtual CPUs (vCPU count) assigned to
each VM guest. Otherwise, VM guests with a vCPU count greater or equal
to WLM's minimum allocation could receive insufficient resources and
eventually crash. For example, if an Integrity VM host has 8 cores and
three guests with 1, 2, and 4 virtual CPUs, respectively, your WLM host
should maintain an allocation of at least 4 cores at all times. You can
achieve this by using the WLM keyword; for more information on this
keyword, see the wlmconf(4) manpage.
To run WLM within an Integrity VM (guest), you cannot use Instant
Capacity, Pay per use, and vPar integration. (However, a guest will
take advantage of CPU resources added to the VM Host by Instant Capac‐
ity, Temporary Instant Capacity, and Pay per use.) As noted previ‐
ously, WLM must continue allocating at least as many cores as the maxi‐
mum number of virtual CPUs in any VM guest on the system. In addition,
specify a WLM interval greater than 60 seconds. This helps ensure a
fair allocation of CPU resources for FSS groups.
For more information on HP Integrity VM, refer to the following web
site and navigate to the "Solution components" page:
www.hp.com/go/vse
AUTHOR
was developed by HP.
FEEDBACK
If you would like to comment on the current HP-UX WLM functionality or
make suggestions for future releases, please send email to:
wlmfeedback@rsn.hp.com
FILES
example WLM configuration file
example WLM global arbiter configuration file
example WLM configuration file
example WLM global arbiter configuration file
system initialization directives
default message log
optional global arbiter statistics log
copy of most recent configuration file
SEE ALSOwlmd(1M), wlmaudit(1M), wlminfo(1M), wlmcert(1M), wlmconf(4), wlmpar‐
conf(4), wlm(5)
HP-UX Workload Manager User's Guide (/opt/wlm/share/doc/WLMug.pdf)
HP-UX Workload Manager homepage (http://www.hp.com/go/wlm)
wlmpard(1M)