HP-UX Workload Manager User's Guide

Example configuration files
transient_groups.wlm
Chapter 9 325
# Set up wlmrcvdc to pick up metrics indicating whether the Serviceguard
# packages are active; set transient_groups keyword.
#
# Have WLM modify allocations (if necessary) every 5 seconds
# because the configuration includes usage goals.
#
tune {
coll_argv = wlmrcvdc sg_pkg_active;
transient_groups = 1;
wlm_interval = 5;
}
# The SLO Paying_slo applies to a PSET-based group, so absolute CPU units
# are in effect. Thus, each core of CPU resources represents 100 shares. The
# mincpu and maxcpu values can be up to (n * 100) shares, where n is the
# total number of cores on the system.
# The SLO for the Apache workload is going to request 1 to 2 cores. It has a
# usage goal, which will ensure the group uses a certain percentage of its
# CPU allocation. This SLO is priority 1, as Apache must get the CPU
# resources it needs.
slo Apache_slo {
pri = 1;
mincpu = 100;
maxcpu = 200;
entity = PRM group Apache;
goal = usage _CPU;
}
# Whenever the Billing Serviceguard package is active, this SLO’s associated
# workload group is allocated 1 to 4 cores, based on a usage goal. When the
# package is not active, the Billing workload group, which is an FSS group,
# is removed from the configuration and consumes no resources. This SLO is
# priority 2; it is not as critical as the Apache workload.
slo Billing_slo {
pri = 2;
mincpu = 100;
maxcpu = 400;
entity = PRM group Billing;
goal = usage _CPU;
condition = metric Billing_active;