Readme file for IBM® Spectrum LSF 10.1 Fix 601993
Abstract
P105112. This fix introduces a new feature to LSF called cluster affinity. When a user submits jobs with the same cluster affinity attribute LSF will run the jobs in the same cluster.
Description
Readme documentation for IBM Spectrum LSF 10.1 Fix 601993 including installation-related instructions, prerequisites and co-requisites, and list of fixes.
This fix introduces a new feature to LSF called cluster affinity. When a user submits jobs with the same cluster affinity attribute LSF will run the jobs in the same cluster. This feature is intended for multicluster environments.
LSF scheduling behaviour
If a job does not have a cluster affinity requirement specified, there
is no change to how the job is scheduled. If a job does have a cluster affinity
requirement specified, but it is not within a multi-cluster environment, there
is no change to how the job is scheduled. If a job does have a cluster affinity
requirement specified and is within a multi-cluster environment, then the
scheduler will use the job’s cluster affinity requirement to determine to which
cluster the job should be forwarded. The first time the scheduler sees a job
with a specific cluster affinity requirement, the job will be scheduled as
normal. LSF will make note of which cluster was used for that specific cluster
affinity requirement. The next time a job with the same specific cluster
affinity requirement requires scheduling, the scheduler will recall the first
job used and forward the new job to the same cluster. Note that if the
configured time-to-live for cluster affinity has elapsed before seeing a new
job with the same cluster affinity requirement, LSF will delete the
corresponding cluster affinity attribute.
Submitting jobs
To submit a job with a cluster affinity attribute, the -jobaff option has been extended to
allow you to specify a cluster affinity attribute. The first time a job is
submitted with a specific cluster affinity attribute, LSF creates the specific
cluster affinity attribute:
bsub -jobaff
“cluster_affinity(attribute_name)” ./a.out
Jobs submitted by the same user with the same cluster affinity attribute name will be forwarded to the same cluster. The scope of the cluster affinity attributes is at the user level within the submission cluster. This means:
• Different users can specify the same cluster affinity attributes, but LSF will treat them as unique.
• Jobs submitted by the same user in different clusters with the same cluster affinity attribute are in a unique namespace based on the submission cluster.
The bmod command can be used to
change a job’s cluster affinity attribute if the job is still pending and has
not been forwarded to a remote cluster.
Viewing cluster affinity attribute information
The battr show command has been extended to display cluster
affinity information:
> battr show
-h
Usage: show {[-json | -w] [-m "host_name/host_group
... "] [-u user_name] [attr_name ...]} | {-cluster_affinity [-w | -l |
-json] [-c cluster_name] [-u user_name | all] [attr_name ...]}
Use the -cluster_affinity
option to view the cluster affinity attributes information:
-cluster_affinity view cluster affinity attribute information
-w wide
format
-l long
format
-json output
JSON format
-c <cluster_name>
show attributes associated with the specified
cluster
-u <user_name> show attributes associated with the
specified user
-u all show
attributes associated with all users
attr_name show attributes with the
specified names
> battr show
-cluster_affinity
NAME TTL(HH:MM:SS) USER
CLUSTER #JOBS DESCRIPTION
tag1 - root - 1 Cluster affinity
attri
tag 00:08:17 root cluster2 0 Cluster affinity
attri
> battr show -cluster_affinity -l
NAME: tag1
TTL(HH:MM:SS): -
USER: user1
CLUSTER: -
DESCRIPTION: Cluster affinity attribute created for job 137.
JOBS: 137
------------------------------------------------------------------------------
NAME: tag
TTL(HH:MM:SS): 00:09:24
USER: user1
CLUSTER: cluster2
DESCRIPTION: Cluster affinity attribute created for job 110.
JOBS:
NAME: Cluster affinity attribute name.
TTL: Time-to-live
if no running or pending jobs associated with the cluster affinity attribute.
USER: The
user.
DESCRIPTION: LSF
generated. The job that caused the creation of this cluster affinity attribute.
If no cluster is associated with a cluster affinity name a hyphen (“-“) will be shown for the CLUSTER field. If there are running or pending jobs associated with a cluster affinity name a hyphen will be shown for the TTL field.
Automatic cluster affinity attribute deletion
LSF will automatically delete
a cluster affinity attribute when time-to-live (TTL) value has expired. The TTL
counter for a cluster affinity attribute begins to decrement when there are no
pending or running jobs that reference it (-jobaff
“cluster_affinity(attribute_name)”). When the counter reaches zero it is
removed. If another job is submitted that references it the TTL counter is
reset.
Clearing cluster associations
Use the battr reset command to clear or reset a cluster affinity attribute association.
This can be useful to deal with the case where a group of jobs get bound to a cluster, but for whatever reason, the whole group cannot complete using the cluster because the cluster goes down. This provides a way to reset,delete, or clear the current cluster association and LSF tries again:
> battr reset -h
Usage: reset -cluster_affinity [-u user_name | all] attr_name
[attr_name ...]
-cluster_affinity reset the cluster affinity association
-u <user_name> reset attributes associated with the
specified user
-u all reset
attributes associated with all users
attr_name reset attributes with the
specified names
> battr reset -cluster_affinity attr1
Cluster affinity attribute <attr1> successfully reset.
The LSF administrator can reset any cluster affinity association. An ordinary user can only reset those that belong to them.
Configuring TTL cluster wide
Configure CLUSTER_AFFINITY_TTL in the lsb.params file. Specifies the time-to-live for newly-created cluster affinity Attributes.
CLUSTER_AFFINITY_TTL=<time_hours>
CLUSTER_AFFINITY_TTL=<time_minutes>m | M
Valid values
Any positive integer between 10 minutes and 100 hours. If this parameter is set to a value that is lower than 10m, CLUSTER_AFFINITY_TTL defaults to the minimum value of 10 minutes. If this parameter is set to a value that is higher than 100, CLUSTER_AFFINITY_TTL defaults to the maximum value of 100 hours.
Default
1 (1 hour)
Readme file for: IBM® Spectrum LSF
Product or component release: 10.1
Update name: Fix 601993
Fix ID: LSF-10.1-build601993
Publication date: 24 April 2024
Contents
1. List of fixes
2. Download location
3. Product or components affected
4. System requirements
5. Installation and configuration
6. List of files
7. Product notifications
8. Copyright and trademark information
1. List of fixes
P105112
2. Download locations
Download Fix 601993 from the following location: https://www.ibm.com/support/fixcentral
3. Product or components affected
Affected product or components include:
LSF/bparams
LSF/bsub
LSF/bmod
LSF/bjobs
LSF/battr
LSF/bhist
LSF/mbatchd
LSF/mbschd
LSF/ebrokerd
LSF/bhosts
LSF/liblsbstream.so
LSF/liblsbstream.a
LSF/libbat.so
LSF/libbat.a
LSF/liblsf.so
LSF/liblsf.a
LSF/lsbatch.h
LSF/lsf.h
LSF/lsfproxyd
4. System requirements
linux2.6-glibc2.3-x86_64
linux3.10-glibc2.17-x86_64
5. Installation and configuration
Before you install
LSF_TOP is the full path to the top-level installation directory of LSF.
1. Before you apply this fix, ensure that you installed LSF 10.1 Fix Pack 14 or above. You can download LSF 10.1 Fix Pack 14 from https://www.ibm.com/support/fixcentral. Search for build601547.
2. Log on to the LSF management host as the LSF primary administrator.
3. Set
your environment:
-For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf
-For sh, ksh,
or bash: $ . LSF_TOP/conf/profile.lsf
Installation steps
1.
Log on to the LSF management host as the root
user and set your environment:
-For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf
-For sh, ksh, or bash: $ . LSF_TOP/conf/profile.lsf
2. Go to the install directory: cd $LSF_ENVDIR/../10.1/install/
3. Copy the fix file to the install directory: $LSF_ENVDIR/../10.1/install/
4. Run patchinstall: ./patchinstall <fix>
After you install
1. Log on to the LSF management host as the LSF primary administrator and set your environment:
-For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf
-For sh, ksh, or bash: $ . LSF_TOP/conf/profile.lsf
2. Run badmin mbdrestart -s
Uninstallation
1. Log on to the LSF management host as the root user and set your environment:
-For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf
-For sh, ksh, or bash: $ . LSF_TOP/conf/profile.lsf
2. Go to the install directory: cd $LSF_ENVDIR/../10.1/install/
3. Run ./patchinstall -r <fix>
4. Log on to the LSF management host as the LSF primary administrator and set your environment:
-For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf
-For sh, ksh, or bash: $ . LSF_TOP/conf/profile.lsf
5. Run badmin mbdrestart -s
6. List of files
The following components in all Linux and UNIX packages:
LSF/bparams
LSF/bsub
LSF/bmod
LSF/bjobs
LSF/battr
LSF/bhist
LSF/mbatchd
LSF/mbschd
LSF/ebrokerd
LSF/bhosts
LSF/liblsbstream.so
LSF/liblsbstream.a
LSF/libbat.so
LSF/libbat.a
LSF/liblsf.so
LSF/liblsf.a
LSF/lsbatch.h
LSF/lsf.h
LSF/lsfproxyd
7. Product Notifications
To receive information about product solution and fix updates automatically, subscribe to product notifications on the My notifications page (www.ibm.com/support/mynotifications) on the IBM Support website (support.ibm.com). You can edit your subscription settings to choose the types of information you want to get notification about, for example, security bulletins, fixes, troubleshooting, and product enhancements or documentation changes.
8. Copyright and Trademark Information
©Copyright IBM Corporation 2024
U.S. Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.
IBM®, the IBM logo, and ibm.com® are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml.