Description of problem: Keystone is configured as a single process on OSP11 / Ocata z5 Version-Release number of selected component (if applicable): OSP11z5 How reproducible: Always Steps to Reproduce: 1. Install Undercloud and see how many Keystone admin/main api processes exist 2. 3. Actual results: Single process despite multi-core system Expected results: Several keystone processes to handle greater load against keystone on undercloud Additional info: Anything > 10 nodes will suffer a slow overcloud install due to the single keystone process only able to use a ~1.2cores while building an overcloud. For Example, we just recently built an 82 node cluster (3 controllers, 10 cephs, 69 computes) and the deploy actually exceeded the 4 hour timeout due to keystone spending most of its time stuck on 1.2 cores. This easily can be tuned to more cores and greatly speeds up the deployment process. OSP10 even has this fix. This was supposedly fixed in OSP11 before according to this bz: https://bugzilla.redhat.com/show_bug.cgi?id=1330980 [root@b04-h19-1029p ~]# cat /etc/version.json { "osp_series": "ocata", "osp_version": "11", "rhos_release": "11-director", "build": "z5", "uc_build_date": "20180501-140811" } [root@b04-h19-1029p ~]# rpm -qa | grep keystone python-keystonemiddleware-4.14.0-1.el7ost.noarch puppet-keystone-10.4.0-1.el7ost.noarch python-keystone-11.0.3-5.el7ost.noarch python-keystoneclient-3.10.0-1.el7ost.noarch openstack-keystone-11.0.3-5.el7ost.noarch python-keystoneauth1-2.18.0-1.el7ost.noarch [root@b04-h19-1029p ~]# ps afx | grep keystone 47316 pts/1 S+ 0:00 \_ grep --color=auto keystone 18547 ? Sl 0:14 \_ keystone-admin -DFOREGROUND 18548 ? Sl 0:09 \_ keystone-main -DFOREGROUND [root@b04-h19-1029p ~]# grep processes /etc/httpd/conf.d/10-keystone_wsgi_* /etc/httpd/conf.d/10-keystone_wsgi_admin.conf: WSGIDaemonProcess keystone_admin display-name=keystone-admin group=keystone processes=1 threads=12 user=keystone /etc/httpd/conf.d/10-keystone_wsgi_main.conf: WSGIDaemonProcess keystone_main display-name=keystone-main group=keystone processes=1 threads=12 user=keystone [root@b04-h19-1029p ~]# lscpu | grep "^CPU(s):" CPU(s): 64 I may have opened this against the incorrect component but I believe this is *fixed* in puppet-keystone but something is passing in the wrong parameters and I am not sure what component that is doing this.
Alex, what's the undercloud.conf that you used to deploy?
Alex, also, you are right, this is still an issue in OSP12/Pike. The fix landed for Queens though.
(In reply to Juan Antonio Osorio from comment #8) > Alex, what's the undercloud.conf that you used to deploy? For OSP11 or OSP12? The OSP11 undercloud.conf is available here: https://gist.github.com/akrzos/c85bd1a281905f88e51c89efeedb84bf
Upstream review has merged. Moving RHBZ to POST.
Downstream build complete. Updatinf FIV and moving to MODIFIED.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:3789