opendev/system-config
Code Issues Proposed changes

Merge "Use bifrost for bare metal portion of infra-cloud"

Browse Source
This commit is contained in:
Jenkins 2015-07-29 18:14:58 +00:00 committed by Gerrit Code Review
commit 5853a14efc
1 changed files with 12 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
doc/source/infra-cloud.rst
View File

@ -92,22 +92,22 @@ Management
* A "Ironic Controller" machine is installed by hand into each site. That * A "Ironic Controller" machine is installed by hand into each site. That
machine is enrolled into the puppet/ansible infrastructure. machine is enrolled into the puppet/ansible infrastructure.
* An all-in-one one node OpenStack cloud with Ironic as the Nova driver * The "Ironic Controller" will have bifrost installed on it. All of the
is installed on each Ironic Controller node. The OpenStack Cloud other machines in that site will be enrolled in the Ironic that bifrost
produced by this installation will be referred to as "Ironic Cloud manages. bifrost will be responsible for booting base OS with IP address
$site". In order to keep things simpler, these do not share anything and ssh key for each machine.
with the cloud that nodepool will make use of.
* Each additional machine in a site will be enrolled into the Ironic Cloud * The machines will all be added to a manual ansible inventory file adjacent
as a bare metal resource. to the dynamic inventory that ansible currently uses to run puppet. Any
metadata that the ansible infrastructure for running puppet needs that
would have come from OpenStack infrastructure will simply be put into
static ansible group_vars.
* Each Ironic Cloud $site will be added to the list of available clouds that * The static inventory should be put into puppet so that it is public, with
launch_node.py or the ansible replacement for it can use to spin up long the IPMI passwords in hiera.
lived servers.
* An OpenStack Cloud with KVM as the hypervisor will be installed using * An OpenStack Cloud with KVM as the hypervisor will be installed using
launch_node and the OpenStack puppet modules as per normal infra OpenStack puppet modules as per normal infra installation of services.
installation of services.
* As with all OpenStack services, metrics will be collected in public * As with all OpenStack services, metrics will be collected in public
cacti and graphite services. The particular metrics are TBD. cacti and graphite services. The particular metrics are TBD.