undercloud heat-engine workers setting inappropriate for large baremetal systems

[Steven Hardy]

https://github.com/openstack/instack-undercloud/blob/master/elements/puppet-stack-config/puppet-stack-config.yaml.template#L144..L147

We have reports (from rook on IRC doing scale testing) that recent releases can use too much memory due to starting #CPUs heat-engine processes.

Since we made that change in instack-undercloud, puppet-* gained the ability to calculate better defaults (which max out at 12 workers) so we should probably switch to those instead.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/529066

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/529066
Committed: https://git.openstack.org/cgit/openstack/instack-undercloud/commit/?id=e95b5d64348832a46d02431ff0fa4244847ad03f
Submitter: Zuul
Branch: master

commit e95b5d64348832a46d02431ff0fa4244847ad03f
Author: Steven Hardy <email address hidden>
Date: Tue Dec 19 13:47:52 2017 +0000

    Set heat engine workers to "%{::os_workers_heat_engine}"

    This should be a better setting for large baremetal environments where
    using the default #CPUs setting can consume too much memory.

    Note the fact added in I31d02bea6dd55d65a7014503398adc4422ce7303 is used,
    this will chose a value similarly to os_workers fact but changing the
    minimum from 2 to 4 as we know 2 doesn't work well for heat/tripleo.

    This is a partial revert of 34ec2644284d59c0e07f2c4729cd354a37928aae

    Change-Id: I0283ef092559fdf9b46c0b04a7d9f4f934ed4fbd
    Closes-Bug: #1739030

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/instack-undercloud 8.2.0 release.