HP-UX Workload Manager User's Guide

Example configuration files
metric_condition.wlm
Chapter 9302
# Purpose:
# Demonstrate the use of HP-UX WLM to enable a service-level objective
# (SLO) when a measurable metric value is reached. Also make use of a
# glance data collector to provide a metric value.
#
# Components:
# The glance toolkit (included with HP-UX WLM) is used. See
# glance_prm(1M) for more information on the glance data collectors.
#
# Dependencies:
# This example was designed to run with version HP-UX WLM A.01.01
# or later.
#
# See wlmconf(4) for complete HP-UX WLM configuration information.
#
#
# prm structure
# Define all workload groups in the prm structure.
#
prm {
groups = OTHERS : 1,
jobs : 2;
apps = jobs: /opt/batch/bin/submit;
}
# slo structures
#
# This SLO has a metric goal. It also has a metric used in the
# condition statement. Both these metrics will require that more
# information be provided in metric-specific tune structures.
#
# This SLO becomes active based on the number of active jobs, once
# the server actually has jobs submitted to it. It attempts to keep
# the average job-completion time under 10 minutes (these are batch
# jobs, so we’re not dealing in seconds of response time) when there
# are any active jobs. If there are no jobs, this SLO is inactive.
#
slo job_processing {
pri = 1; # SLO of highest priority
mincpu = 40; # minimum CPU allocation (percentage)
maxcpu = 90; # maximum CPU allocation (percentage)
entity = PRM group jobs;
goal = metric avg_completion_time < 10;
condition = metric number_of_active_jobs > 0;