ee5f25a726
By default many services use the number of cpus as the default number of workers. On baremetal systems this can lead to excessive RAM comsumption as this can be 16, 32, 48 or higher for each of the services. In the upstream puppet openstack, a fact was created as part of the Newton cycle that is the larger value between (number of cpus/4) or 2 but is capped at 8. These defaults have not been updated yet in the upstream modules and it would be better to specifically call out the worker configurations in our puppet-stack-config.yaml template. This change updates the worker settings for the classes that are pulled in for the undercloud to leverage this tuned fact to lower the overall number of workers for all the services. This should improve memory consumption and reduce the number of processes running on baremetal systems. The os_workers fact will be 2 for unless the cpu count is greater than 8 with an incremental increase of 1 worker for every 4 processors until 32 processors. Anything above 32 processors will still result in a worker value of 8. This change updates 17 services that currently may use the number of processors for the default number of workers. Change-Id: I9ed855648e23b0a7e452e6a840a92779fa3f6d48 Closes-Bug: #1633544 |
||
|---|---|---|
| .. | ||
| extra-data.d | ||
| install.d | ||
| os-apply-config/etc/puppet/hieradata | ||
| os-refresh-config | ||
| README.rst | ||
| element-deps | ||
| package-installs.yaml | ||
| puppet-stack-config.pp | ||
| puppet-stack-config.yaml.template | ||
README.rst
puppet-stack-config
puppet-stack-config provides static puppet configuration for a single node baremetal cloud using the Ironic driver. A yaml template is used to render a hiera data file at /etc/puppet/hieradata/puppet-stack-config.yaml.
The template rendering takes its input from a set of defined environment variables.