bgpinfo

Displays information about global fairshare and global resources.

Synopsis

bgpinfo subcommand
bgpinfo [-h | -V]

Description

The bgpinfo command provides a set of subcommands to get information about global fairshare.

This command can also display remote fairshare load. The remote fairshare load impacts your dynamic priority for job scheduling.

Information about each subcommand is available through the help command.

Subcommand synopsis

conf [-l] [global_policy_name...]
fsload [-l [-c cluster_name]] [global_policy_name...]
policy
resource [-c] [-l] [-s] [-w]
status [-l [global_policy_name ...]]
-h
-V

Options

subcommand

Runs the specified subcommand. See Usage section.

-h

Prints command usage to stderr and exits.

-V

Prints LSF release version to stderr and exits.

Usage

conf [-l] [global_policy_name ...]
Shows summary of global fairshare policy configurations.
POLICY_NAME
The name of the global policy.
TYPE
Type of global policy.

If global_policy_name is specified, shows only the information for the specified global policies.

-l
Long format. Shows the following additional information:
SYNC_INTERVAL
Fairshare load synchronization interval.
SYNC_MODE
Synchronization mode (all or partial).
PARTICIPANTS
Participants in the global fairshare policy.
$ bgpinfo conf
POLICY_NAME       TYPE
low               fairshare
admin             fairshare
 
$ bgpinfo conf -l
POLICY NAME: low
-- A global fairshare policy for low queue.
  TYPE: fairshare
  PARAMETERS:
  SYNC_INTERVAL: 30 seconds
  SYNC_MODE: partial
  PARTICIPANTS: low@c1 low@c2
 
POLICY NAME: admin
-- A global fairshare policy for admin queue.
  TYPE: fairshare
  PARAMETERS:
  SYNC_INTERVAL: 30 seconds
  SYNC_MODE: all
  PARTICIPANTS: admin@c1 admin@c2
 
$ bgpinfo conf -l admin
POLICY NAME: admin
-- A global fairshare policy for admin queue.
  TYPE: fairshare
  PARAMETERS:
  SYNC_INTERVAL: 30 seconds
  SYNC_MODE: all
  PARTICIPANTS: admin@c1 admin@c2
fsload [-l [-c cluster_name]] [global_policy_name ...]
Shows the aggregated fairshare load for global fairshare policies.

By default, displays recursively the summed fairshare load information for the global share account tree for all global fairshare policies. Use the fsload -l option to display fairshare load information for each participating queue. Only share accounts whose fairshare load entries are not all zero are shown.

The POLICY_NAME is the name of the global fairshare policy. The following information is displayed for fairshare load:
SHARE_INFO_FOR
The sum of the fairshare load information from all participating queues, users, or user groups in the global fairshare policy.
USER/GROUP
Names of users or user groups who have access to the global fairshare policy.
STARTED
Number of job slots that are used by running or suspended jobs from users or user groups that participate in the global fairshare policy.
RESERVED
Number of job slots that are reserved by jobs from users or user groups that participate in the global fairshare policy.
CPU_TIME
Cumulative CPU time that is used by jobs from users or user groups that participate in the global fairshare policy.
RUN_TIME
Total run time of all participants in the global fairshare policy.
HIST_RUN_TIME
Total historical run time of all participants in the global fairshare policy.
ADJUST
Total dynamic priority calculation adjustment for all participants in the global fairshare policy.
STARTED_JOBS
The number of started jobs and suspended jobs as used by the fairshare scheduling algorithm. This field only displays if FAIRSHARE_JOB_COUNT=Y is enabled in the lsb.params file.
RESERVED_JOBS
The number of reserved jobs as used by the fairshare scheduling algorithm. This field only displays if FAIRSHARE_JOB_COUNT=Y is enabled in the lsb.params file.
-l
In addition to default output, shows fairshare load information for each participant.
The PROVIDER and SHARE_INFO_FOR show fairshare load information that is provided by a specific participant:
USER/GROUP
Name of users or user groups who have access to the global fairshare policy.
STARTED
Number of job slots that are used by running or suspended jobs from users or user groups for the specific participant.
RESERVED
Number of job slots that are reserved by jobs from users or user groups for the specific participant.
CPU_TIME
Cumulative CPU time that is used by jobs from users or user groups for the specific participant.
RUN_TIME
Run time for the specific participant.
ADJUST
Dynamic priority calculation adjustment for the specific participant.
-c cluster_name
Filter for the -l option. It shows fairshare load for the specified participating cluster. Only one cluster name can be specified with the -c option.
$ bgpinfo fsload -l -c c1 admin
POLICY_NAME: admin

  SHARE_INFO_FOR: /
  USER/GROUP   STARTED  RESERVED  CPU_TIME  RUN_TIME  HIST_RUN_TIME   ADJUST
  ugroup1            27        0       2.9    262464              0    0.000

    PROVIDER: admin@c1
    USER/GROUP   STARTED  RESERVED  CPU_TIME  RUN_TIME  HIST_RUN_TIME   ADJUST
    ugroup1           18         0       2.9     92129              0    0.000

  SHARE_INFO_FOR: /ugroup1/
  USER/GROUP   STARTED  RESERVED  CPU_TIME  RUN_TIME  HIST_RUN_TIME   ADJUST
  user1              9         0       1.0     87490              0     0.000
  user2              9         0       0.0     87488              0     0.000
  user3              9         0       1.0     87486              0     0.000

    PROVIDER: admin@c1
    USER/GROUP   STARTED  RESERVED  CPU_TIME  RUN_TIME  HIST_RUN_TIME   ADJUST
    user1              6         0       1.0     30711              0    0.000
    user2              6         0       0.9     30709              0    0.000
    user3              6         0       1.0     30709              0    0.000
policy
Shows distribution policies for global resources and global limits.
resource [-c] [-l] [-s] [-w]
Shows usage information on global resources.
-c
Shows the global resource configuration information instead of usage information.
-l
Shows detailed usage information on global resources for all clusters.
-s
Shows filtered usage information on global resources.
-w
Shows usage information on global resources in wide format.
status [-l [global_policy_name ...]]
Shows the status of the gpolicyd daemon and global fairshare policies. By default, the following information is shown:
GPOLICYD CLUSTER
The cluster where gpolicyd is running.
GPOLICYD HOST
The host where gpolicyd is running.
-l
In addition to the default output for daemon gpolicyd status, shows the status of specified global fairshare policies. By default, shows status of all global fairshare policies.
GLOBAL FAIRSHARE POLICY shows the status of a global fairshare policy:
POLICY NAME
Name of the global fairshare policy.
PARTICIPANTS
Name of participants of the global fairshare policy.
STATUS
Status of the participant.
  • Disconnected - The cluster of the participant is disconnected from gpolicyd
  • Rejected - Registration rejected. The participant does not exist or the participant is not a fairshare queue.
  • ok - The cluster for the participant is connected to gpolicyd and the queue for the participant is successfully registered with gpolicyd.
$ bgpinfo status 
GPOLICYD HOST: userA
GPOLICYD CLUSTER: c1 

$ bgpinfo status -l 
GPOLICYD HOST: userA
GPOLICYD CLUSTER: c1 

GLOBAL FAIRSHARE POLICY 

POLICY NAME: low 
  PARTICIPANTS         STATUS 
  low@c1               ok     
  low@c2               rejected   

POLICY NAME: admin 
  PARTICIPANTS         STATUS 
  admin@c1             ok
  admin@c2             ok