Retiring a node in Ironic

Overview

Retiring nodes is a natural part of a server’s life cycle, for instance when the end of the warranty is reached and the physical space is needed for new deliveries to install replacement capacity.

However, depending on the type of the deployment, removing nodes from service can be a full workflow by itself as it may include steps like moving applications to other hosts, cleaning sensitive data from disks or the BMC, or tracking the dismantling of servers from their racks.

Ironic provides some means to support such workflows by allowing to tag nodes as retired which will prevent any further scheduling of instances, but will still allow for other operations, such as cleaning, to happen (this marks an important difference to nodes which have the maintenance flag set).

Requirements

The use of the retirement feature requires that automated cleaning be enabled. The default conductor.automated_clean setting must not be disabled as the retirement feature is only engaged upon the completion of cleaning as it sets forth the expectation of removing sensitive data from a node.

If you’re uncomfortable with full cleaning, but want to make use of the the retirement feature, a compromise may be to explore use of metadata erasure, however this will leave additional data on disk which you may wish to erase completely. Please consult the configuration for the deploy.erase_devices_metadata_priority and deploy.erase_devices_priority settings, and do note that clean steps can be manually invoked through manual cleaning should you wish to trigger the erase_devices clean step to completely wipe all data from storage devices. Alternatively, automated cleaning can also be enabled on an individual node level using the baremetal node set --automated-clean <node_id> command.

How to use

When it is known that a node shall be retired, set the retired flag on the node with:

baremetal node set --retired node-001

This can be done irrespective of the state the node is in, so in particular while the node is active.

Note

An exception are nodes which are in available. For backwards compatibility reasons, these nodes need to be moved to manageable first. Trying to set the retired flag for available nodes will result in an error.

Optionally, a reason can be specified when a node is retired, e.g.:

baremetal node set --retired node-001 \
  --retired-reason "End of warranty for delivery abc123"

Upon instance deletion, an active node with the retired flag set will not move to available, but to manageable. The node will hence not be eligible for scheduling of new instances.

Equally, nodes with retired set to True cannot move from manageable to available: the provide verb is blocked. This is to prevent accidental reuse of nodes tagged for removal from the fleet. In order to move these nodes to available none the less, the retired field needs to be removed first. This can be done via:

baremetal node unset --retired node-001

In order to facilitate the identification of nodes marked for retirement, e.g. by other teams, ironic also allows to list all nodes which have the retired flag set:

baremetal node list --retired