A spare node is a standby node that is registered and managed in units of a fault domain.
You can register one spare node for each fault domain.
When maintenance recovery of the auto-recovery function cannot recover storage nodes from a failure, the faulty storage node switches to a spare node in the same fault domain, and then the spare node becomes a primary storage node.
- Server models other than those described in the VSP One SDS Block Bare Metal Hardware Compatibility Reference do not support the spare node function, and therefore might not allow for spare node switchover. For details about server models that support spare node switchover, see the VSP One SDS Block Bare Metal Hardware Compatibility Reference.
- If communication with the storage node and the BMC is not possible, spare node switchover will be unsuccessful. When using the spare node function, make VSP One SDS Block configuration redundant to prevent communication with the storage node and the BMC from being disconnected due to a single power failure or a single network failure.
The replaced storage node is removed from a storage cluster.
Using a spare node function enables quick recovery of degraded redundancy.
If you cannot prepare the same number of spare nodes as fault domains, place spare nodes in fault domains that contain many storage nodes with a clusterRole value of “Master”.
The following shows an example of a spare node configuration when the number of fault domains is three.
In the configuration example, a spare node is installed in each fault domain.
If auto recovery for a storage node failure in a fault domain fails while the spare node function is enabled, the storage node is switched to the spare node within the fault domain. The following figures show the flow of switching a storage node to a spare node.
-
Failure occurred in a storage node and the node cannot be recovered by auto recovery.
-
Switch the faulty storage node to a spare node.
-
The OS of the storage node that has a failure is shut down. After spare node switchover completes, the spare node operates as a storage node.
When you use the spare node function, you need to make the control port of the storage node and the control port of the spare node belong to the same subnet. This also applies to the compute port of the storage node and the compute port of the spare node.
Also, you can set multiple compute ports in one storage node and make them belong to different subnets. In this case, you need to make the compute network of the spare node (to which the storage node is to switch) belong to the same subnet as the subnet of the compute network of the storage node.
The following figure shows the relationship between two compute ports that are set for one storage node and belong to separate subnets and those of the spare node that takes over the compute network.
In the preceding example, control ports of the storage node and the spare node belong to the subnet of 192.168.0.*/24. When you set two compute ports in the storage node and make the compute ports belong to the subnet of 192.168.10.*/24 and 192.168.11.*/24 respectively, the compute ports of the spare node that inherits the network information must belong to the same subnets.
For details, see Managing spare nodes (Bare metal) in the VSP One SDS Block and SDS Cloud System Administration.