Note:
Components that are hot-pluggable do not require the server to be powered down.
For more information, see the component replacement procedures in this guide.
Before Powering Down a Node
Servers in a Private Cloud Appliance rack are integrated components of a multinode
environment. Unless a server experiences a sudden failure or must be shut down in an
emergency situation, powering down server nodes in the appliance should be planned and
performed carefully to minimize impact on the performance and availability of the appliance
services and resources.
When planning to power down server nodes, take these points into account:
System load
Powering down servers means system capacity is temporarily reduced. It makes sense to plan
maintenance operations at a time when system load is typically at its lowest; for example
during the night, over the weekend, and so on.
Instance availability
Many workloads need to remain available while a server is offline for maintenance. You can
live-migrate compute instances to other compute nodes in the same fault domain, or restart
them in another fault domain. The available compute capacity in the three fault domains
determines whether all workloads can remain operational or if some need to be scaled down
or even interrupted temporarily.
Maintenance mode
Before you can safely take a server offline, it must be placed in maintenance mode. In a
Private Cloud Appliance environment this means that a compute node must not host any
running compute instances, that it is under maintenance lock, and that provisioning has been
disabled. An appliance administrator must perform these tasks from the Service Enclave.
For more information about locking nodes for maintenance, migrating compute instances, and
configuring high availability in the Compute service, refer to the chapter Hardware
Administration in the Oracle Private Cloud Appliance Administrator Guide.
Power Down a Node from the Service Enclave
Note:
Always perform a graceful shutdown. It ensures that all data is saved and the system
is ready for restart.
1. Ensure that no compute instances are running on this node, and that the maintenance and
provisioning locks are active.
2. From the Service Enclave, issue the Stop command.
See Hardware Administration in the Oracle Private Cloud Appliance Administrator Guide.
• Service Web UI: In the Rack Units table, click the action menu (three vertical dots) and
select Stop.
Chapter 4
Preparing the Server for Component Replacement
4-5
To Next Page
Loading...
