lsload

Displays load information for hosts.

Synopsis

lsload [-a] [-cname] [-gpu] [-gpuload] [-l | -w | -o "field_name[:[-][output_width]] ... [delimiter='character']" [-json]] [-N | -E] [-I load_index[:load_index] ...] [-n num_hosts] [-R res_req] [host_name ... | host_group ... | cluster_name ...]
lsload -s [resource_name ...] [-cname] [-loc]
lsload [-h | -V]

Description

Load information can be displayed on a per-host basis, or on a per-resource basis.

By default, displays load information for all hosts in the local cluster.

With the IBM Spectrum LSF multicluster capability enabled, the lsload command also displays load information for all hosts in equivalent clusters.

By default, displays raw load indices.

By default, load information for resources is displayed according to CPU and paging load.

Options

-a
Dynamic Cluster only. Shows information about all hosts, including Dynamic Cluster virtual machine hosts configured with the jobvm resource. Default output includes only standard LSF hosts and Dynamic Cluster hosts configured with the dchost resource.
-cname
In LSF Advanced Edition, includes the cluster name for execution cluster hosts and host groups in output.
Note: This command option is deprecated and might be removed in a future version of LSF.
-gpu
Displays host-based GPU information.
-gpuload
Displays GPU-based information.
-json
Displays the customized output in JSON format.

When specified with the -o option, displays the customized output in the JSON format.

This option should be used with the -o option. If this option is not used with the -o option, the LSF_LSLOAD_FORMAT parameter in the lsf.conf file or the LSF_LSLOAD_FORMAT environment variable must be defined.

-l
Long format. Displays load information without truncation along with extra fields for I/O and external load indices.

This option overrides the index names that are specified with the -I option.

This option does not display information about GPU resources (that is, this option does not display gpu_<num>n resources). Use the -gpu option to view host-based GPU information or the -gpuload option to view GPU-based information.

-N
Displays normalized CPU run queue length load indices.
-E
Displays effective CPU run queue length load indices. Options -N and -E are mutually exclusive.
-w
Displays load information in wide format. Fields are displayed without truncation.
-I load_index[:load_index] ...
Displays only the specified load indices. Separate multiple index names with colons (for example, r1m:pg:ut).

Specify any built-in load index. Specify external load indices only for host-based resources that are numeric and dynamic (you cannot specify external load indices for shared, string, or Boolean resources).

-n num_hosts
Displays only load information for the requested number of hosts. Information for up to num_hosts hosts that best satisfy the resource requirements is displayed.
-o

Sets the customized output format.

  • Specify which lsload fields, in which order, and with what width to display.
  • Specify the asterisk wildcard character (*) in the field name to specify multiple external resource names. You can only specify one asterisk, but this asterisk can be at any position in the field name.

    For example, running lsload -o "gpu_mode*" shows fields such as gpu_mode0, gpu_mode1, gpu_mode2, gpu_model0, gpu_model1, and gpu_model2.

  • Specify only the lsload field name to set its output to unlimited width and left justification.
  • Specify the width colon (:) without a width to set the output width to the recommended width for that field.
  • Specify the width colon (:) with a width to set the maximum number of characters to display for the field. When its value exceeds this width, lsload truncates the ending characters.
  • Specify a hyphen (-) to set right justification when lsload displays the output for the specific field. If not specified, the default is to set left justification when lsload displays the output for a field.
  • Specify the unit colon (:) with a unit to set the unit for the output of the specific field:
    • Specify S to use a built-in conversion for space or capacity, such as memory or disk space. Values are automatically scaled for M (MB), G (GB), and T (TB), where the default unit is M (MB).
      For example, when displaying the mem field with a specified width of 3,
      • For a value of 30, running the lsload -o "mem:3:S" command shows 30.0M.
      • For a value of 4096, running the lsload -o "mem:3:S" command shows 4.0G.
      • For a value of 5000000, running the lsload -o "mem:3:S" command shows 4.8T.
    • Specify D to use a built-in conversion for duration or time, such as memory or disk space. Values are automatically scaled for s (seconds), m (minutes), h (hours), and d (days), where the default unit is s (seconds). The automatically scaled value is rounded up after the first decimal point.
      For example, when displaying the external mytime resource field with a specified width of 5,
      • For a value of 30, running the lsload -o "mytime:5:D" command shows 30.0s.
      • For a value of 8000, running the lsload -o "mytime:5:D" command shows 2.2h.
      • For a value of 5000000, running the lsload -o "mytime:5:D" command shows 57.8d.
    • Specify any other string of 1 - 3 characters and the characters are used as is in the field value. The first character must be a letter (upper or lower case). The second and third characters must be an alphanumeric character.

      For example, when displaying the external gpu_temp resource with a width of 3, running the lsload -o "gpu_temp:3:C" command for a value of 30 shows 30C

  • Use delimiter= to set the delimiting character to display between different headers and fields. This delimiter must be a single character. By default, the delimiter is a space.
Output customization applies only to the output for certain lsload options:
  • LSF_LSLOAD_FORMAT and lsload -o both apply to output for the lsload command with no options, and for lsload options with short form output that filter information, including the following options: -a, -E, -cname, -N, -n, -R.
  • LSF_LSLOAD_FORMAT and lsload -o do not apply to output for lsload options that use a modified format, including the following options: -I, -l, -w, -s.

The lsload -o option overrides the LSF_LSLOAD_FORMAT environment variable, which overrides the LSF_LSLOAD_FORMAT setting in lsf.conf.

By default, the lsload command displays the built-in resource indices. You can also specify the names of external resources. The following are the field names for the built-in resource indices that are used to specify the lsload fields to display, recommended width, and units of measurement for the displayed field:

Table 1. Output fields for lsload
Field name Width Unit
HOST_NAME 20  
status 15  
r15s 6  
r1m 6  
r15m 6  
ut 6  
pg 6  
ls 6  
it 6  
io 6  
tmp 10 LSF_UNIT_FOR_LIMITS in lsf.conf (KB by default)
swp 10 LSF_UNIT_FOR_LIMITS in lsf.conf (KB by default)
mem 10 LSF_UNIT_FOR_LIMITS in lsf.conf (KB by default)
gpu_status*

For example, gpu_status0 and gpu_status1 if there are two GPUs.

10  
gpu_error*

For example, gpu_error0 and gpu_error1 if there are two GPUs.

20  

Field names are case-sensitive. Valid values for the output width are any positive integer 1 - 4096.

For example,

lsload -o "HOST_NAME status: r15s:- r1m:7 r15m:-8 tmp:S swp::S mem:9:S delimiter='^'"

This command displays the following fields:

  • HOST_NAME with unlimited width and left-aligned.
  • status with a maximum width of 15 characters (which is the recommended width) and left-aligned.
  • r15s with a maximum width of 6 characters (which is the recommended width) and right-aligned.
  • r1m with a maximum width of 7 characters and left-aligned.
  • r15m with a maximum width of 8 characters and right-aligned.
  • tmp with unlimited width, left-aligned, and automatically scaled for space or capacity (MB, GB, and TB).
  • swp with a maximum width of 10 characters (which is the recommended width), left-aligned, and automatically scaled for space or capacity (MB, GB, and TB)
  • mem with a maximum width of 9 characters, left-aligned, and automatically scaled for space or capacity (MB, GB, and TB)
  • The ^ character is displayed between different headers and fields.
-R res_req
Displays only load information for hosts that satisfy the specified resource requirements.

Load information for the hosts is sorted according to load on the specified resources.

If res_req contains special resource names, only load information for hosts that provide these resources is displayed (run lshosts to find out what resources are available on each host).

If one or more host names are specified, only load information about the hosts that satisfy the resource requirements is displayed.

With the IBM Spectrum LSF multicluster capability, when a cluster name is specified, displays load information of hosts in the specified cluster that satisfy the resource requirements.

host_name ... | host_group ... | cluster_name ...
Displays only load information for the specified hosts, host groups, or a combination; or clusters.

To display host group information, the LSF_HOSTGROUP_INFO=Y setting must be specified in the lsf.conf file.

With the IBM Spectrum LSF multicluster capability, displays only load information for hosts in the specified clusters.

-s [resource_name ...] [-loc]
Displays information about all dynamic resources that are configured in the cluster, or about the specified resources only. Specify dynamic resources (shared or host-based).

This option does not display information about GPU resources (that is, this option does not display gpu_<num>n resources). Use the -gpu option to view host-based GPU information or the -gpuload option to view GPU-based information.

If the LOCATION parameter in the lsf.cluster.clustername file is set to all to indicate that the resource is shared by all hosts in the cluster, the LOCATION field in the lsload -s command output also displays ALL. To display the individual names of all the hosts in the cluster in the lsload -s command output, specify the -loc option together with the -s option.

-h
Prints command usage to stderr and exits.
-V

Prints LSF release version to stderr and exits.

Default host-based output

Built-in load indices include r15s, r1m, r15m, ut, pg, io, ls, it, swp, mem, and tmp. External load indices are configured in the file lsf.cluster.cluster_name. The selection and order sections of resource requirements control for which hosts are displayed and how the information is ordered.

The display includes the following fields:
HOST_NAME
Standard host name that is used by LSF, typically an internet domain name with two components.
status
Status of the host. A minus sign (-) can precede the status, indicating that RES is not running on the host.
The following statuses are displayed:
ok
The host is in normal state and can accept remote jobs. The ok status indicates that the Load Information Manager (LIM) is unlocked and that both LIM and the Remote Execution Server (RES) are running.
-ok
The LIM on the host is running but RES is unreachable.
busy
The host is overloaded because some load indices exceed configured thresholds. Load index values that caused the host to be busy are preceded by an asterisk (*).
lockW
The host is locked by its run window. Run windows for a host are specified in the lsf.conf configuration file, and can be displayed by the lshosts command. A locked host does not accept LSF jobs from other hosts.
lockU
The host is locked by the LSF administrator or root.
unavail
The host is down or the LIM on the host is not running.
r15s
The 15 second exponentially averaged CPU run queue length.
r1m
The 1 minute exponentially averaged CPU run queue length.
r15m
The 15 minute exponentially averaged CPU run queue length.
ut
The CPU utilization exponentially averaged over the last minute, 0 - 1.
io
By default, io is not shown.

If the -l option is specified, shows the disk I/O rate exponentially averaged over the last minute, in KB per second.

pg
The memory paging rate exponentially averaged over the last minute, in pages per second.
ls
The number of current login users.
it
On UNIX, the idle time of the host (keyboard is not touched on all logged in sessions), in minutes.

On Windows, the it index is based on the time that a screen saver is active on a particular host.

tmp
The amount of free space in /tmp, in MB.
swp
The amount of available swap space.

By default, the amount is displayed in KB. The amount can appear in MB depending on the actual system swap space. Use the LSF_UNIT_FOR_LIMITS parameter in the lsf.conf file to specify a larger unit for the limit (GB, TB, PB, or EB).

mem
The amount of available RAM.

By default, the amount is displayed in KB. The amount can appear in MB depending on the actual system memory. Use the LSF_UNIT_FOR_LIMITS parameter in the lsf.conf file to specify a larger unit for the limit (GB, TB, PB, or EB).

external_index
By default, external load indices are not shown.

If the -l option is specified, shows indices for all dynamic custom resources available on the host, including shared, string, and Boolean resources.

If the -I load_index option is specified, shows only indices for specified non-shared (host-based) dynamic numeric custom resources.

Resource-based output with lsload -s

Displays information about shared and host-based dynamic resources. Each line gives the value and the associated hosts for an instance of the resource.
The displayed information consists of the following fields:
RESOURCE
Name of the resource.
VALUE
Value for an instance of the resource.
LOCATION
Hosts associated with the instance of the resource.

Examples

The following command displays the load of ALPHA hosts with at least 20 MB of swap space, and a 1-minute run queue length less than 0.5.

lsload -R "select[r1m<=0.5 && swp>=20 && type==ALPHA]"

The following command specifies the same resource requirements in restricted format:

lsload -R r1m=0.5:swp=20:type=ALPHA

The following command displays the load of the hosts whose swap space utilization is less than 75%. The resulting hosts are ordered by paging rate.

lsload -R "select[(1-swp/maxswp)<0.75] order[pg]"

The following command displays the 1-minute CPU raw run queue length, the CPU utilization, the disk I/O, and paging rates for all hosts in the cluster.

lsload -I r1m:ut:io:pg

The following command displays the load of all hosts, ordered by r15s:pg. The CPU run queue lengths are the effective run queue lengths.

lsload -E

Diagnostics

Exit status is -10 for LSF problems or an invalid resource names.

Exit status is -1 if an invalid option is specified.

Normal exit status for the lsload command is 0.

See also

lim, lsf.cluster, lsplace, lshosts, lsinfo, lslockhost, ls_load.