HostMaintenancePolicy contains the maintenance policy for the hosts on which the GKE VMs run on.
JSON representation |
---|
{ "maintenanceInterval": enum ( |
Fields | |
---|---|
maintenanceInterval |
Specifies the frequency of planned maintenance events. |
Union field maintenance_strategy . Set of host maintenance strategies available to the customer, all require the maintenance_interval to be PERIODIC. If no strategy is set, and the interval is periodic, customer will be expected to trigger maintenance manually or let maintenance trigger at its initial scheduled time. maintenance_strategy can be only one of the following: |
|
opportunisticMaintenanceStrategy |
Strategy that will trigger maintenance on behalf of the customer. |
MaintenanceInterval
Allows selecting how infrastructure upgrades should be applied to the cluster or node pool.
Enums | |
---|---|
MAINTENANCE_INTERVAL_UNSPECIFIED |
The maintenance interval is not explicitly specified. |
AS_NEEDED |
Nodes are eligible to receive infrastructure and hypervisor updates as they become available. This may result in more maintenance operations (live migrations or terminations) for the node than the PERIODIC option. |
PERIODIC |
Nodes receive infrastructure and hypervisor updates on a periodic basis, minimizing the number of maintenance operations (live migrations or terminations) on an individual VM. This may mean underlying VMs will take longer to receive an update than if it was configured for AS_NEEDED. Security updates will still be applied as soon as they are available. |
OpportunisticMaintenanceStrategy
Strategy that will trigger maintenance on behalf of the customer.
JSON representation |
---|
{ "nodeIdleTimeWindow": string, "maintenanceAvailabilityWindow": string, "minNodesPerPool": string } |
Fields | |
---|---|
nodeIdleTimeWindow |
The amount of time that a node can remain idle (no customer owned workloads running), before triggering maintenance. A duration in seconds with up to nine fractional digits, ending with ' |
maintenanceAvailabilityWindow |
The window of time that opportunistic maintenance can run. Example: A setting of 14 days implies that opportunistic maintenance can only be ran in the 2 weeks leading up to the scheduled maintenance date. Setting 28 days allows opportunistic maintenance to run at any time in the scheduled maintenance window (all A duration in seconds with up to nine fractional digits, ending with ' |
minNodesPerPool |
The minimum nodes required to be available in a pool. Blocks maintenance if it would cause the number of running nodes to dip below this value. |