universal-application-multi-site-manager-demo-cluster¶
CentOS Tasks to do during kickstart install
This pipeline creates a four region edge sites for testing Multi-Site manager functionality. It provides a quick and automated way to build a working distributed system that follows RackN guidelines for a distributed system including having unique management users.
The multi-site system will use DRP endpoint names based on the cluster name. Since these must be registered in your DRP license, we recommend reusing the same cluster names during testing.
Users are highly encouraged to set the dr-server/initial-password
param for security purposes (default is r0cketsk8ts)
The process is as follows:
- cluster calls broker to build 4 machine cluster (cluster/count is ignored)
- new machines start the DRP server install pipeline
- dr-install is installed using
bootstrap/install.sh
or downloaded - dr-install will run with --universal flag set to bootstrap
- cluster waits for all sites to be up, then will register (and remove) Endpoints from manager. This is performed by the registered flexiflow tasks in the pipeline.
This process will create TWO entries for each edge site: one for the machine created by the broker that runs the dr-install process and a second for the self-runner on the edge site. The self-runner is actually registered and running from the edge site and visible because of the management mirroring.
After the DRP endpoints are registered, operations should add
set the universal-stable
or universal-tip
version sets for the Endpoints
and enable synchronization.
To demostrate functionality, operators may create a cluster on the Manager
using the context-broker
and select one of the registered Endpoints. This
will create container backed machines on the edge endpoint that are also mirrored
in the Manager for control.
Note
It is primarily tested with the Linode broker but is not cloud specific. In Linode, the automation will explicitly use multiple regions. To duplicate this effect when using other clouds, make a copy of this pipeline and provide cloud specific region/image information.