Description of problem: Starting Ocata upstream moved to more sane defaults (https://github.com/openstack/puppet-openstacklib/blob/master/lib/facter/os_workers.rb) as deploying with $::processorcount leads to unexpected memory growth/database problems due to each worker opening connections to the database. On a 56 core machine, we should be seeing 12 workers as the default is to use min(processorcount/2, 12). When deploying using default we see api_workers and rpc_workers set empty string and 112 workers processes are seen for neutron server (56 for api and 56 for rpc). However, starting OSP11 we should only see 24 workers in this case (12 for api and 12 for rpc). Version-Release number of selected component (if applicable): 11 How reproducible: 100% Steps to Reproduce: 1. Deploy overcloud with defauls 2. 3. Actual results: 56 worker processes each for neutron api and neutron rpc are seen Expected results: 12 workers processes each for neutron api and neutron rpc should be seen. Additional info:
Might need a backport of https://review.openstack.org/#/c/425423/ which updated the logic to fall back to the os_workers provided by puppet modules if THT is unset
Backport in progress.
This bugzilla has been removed from the release and needs to be reviewed and Triaged for another Target Release.
*** Bug 1470363 has been marked as a duplicate of this bug. ***
OSP11 is now retired, see details at https://access.redhat.com/errata/product/191/ver=11/rhel---7/x86_64/RHBA-2018:1828