Cluster Control APIs
The cluster control APIs provide configuration, activation, and management
functions for the cluster and nodes in the cluster.
For additional information, see:
The cluster control APIs are:
When a partition is detected, some APIs cannot be run in any of the partitions
and some other APIs may be run in any partition. However, the action performed
by the API will take effect only in the partition running the API. The
restrictions for each API are:
- Add Cluster Node Entry
- Not allowed in any partition.
- Add Device Domain Entry
- Only allowed for existing device domain where all members are in the same
partition.
- Adjust Cluster Version
- Not allowed in any partition.
- Change Cluster Node Entry
- To change cluster interface addresses, allowed only within the same
partition. To change node status, allowed only in partition containing the
non-failed nodes.
- Change Cluster Resource Services
- Allowed in any partition.
- Create Cluster
- Not allowed in any partition.
- Delete Cluster
- Allowed in any partition.
- End Cluster Node
- Allowed within the same partition as the node being ended.
- List Cluster Information
- Allowed in any partition.
- List Device Domain Information
- Allowed in any partition.
- Remove Cluster Node Entry
- Allowed in any partition.
- Remove Device Domain Entry
- Only allowed if all members are in the same partition.
- Retrieve Cluster Information
- Allowed in any partition.
- Retrieve Cluster Resource Services Information
- Allowed in any partition.
- Start Cluster Node
- Allowed in any partition.
A network attribute is used to control cluster access within a network. The
network attribute is ALWADDCLU (Allow Add to Cluster). This attribute can be
set as follows:
- *NONE - The system cannot be added to any cluster.
- *ANY - The system will be added to a cluster without verification.
- *RQSAUT - The system will be added to a cluster pending successful
verification through digital certificate exchange.
The default value is *NONE. If *RQSAUT is specified,
see
SSL for software which must be installed on the systems.
Each cluster node has a status associated with it. The status of a cluster
node may govern the behavior of a particular API call. See the individual API
descriptions for more details. The possible values are:
- 1
- New. A node has been added to the cluster membership list
but the Cluster Resource Services has never been started on that node. The
Cluster Resource Service data structures have not been created on the node.
During a Create Cluster operation, the Cluster Resource Service data structures
will be created only on the node running the Create Cluster API.
- 2
- Active. The node has been started either with the Create
Cluster API or Add Cluster Node Entry API with the "Start indicator" parameter
set to 1, or with the Start Cluster Node API. Cluster Resource Services is
active on the node .
- 3
- Remove Pending. The node is in the process of being
removed from the cluster membership list as the result of a Remove Cluster Node
Entry API.
- 4
- Active Pending. The node is in the process of being
started either as the result of a Create Cluster API or Add Cluster Node Entry
API call with the "Start indicator" parameter set to 1 or because of a Start
Cluster Node API call. In addition, the node could have previously had a status
of Partition and will change to the Active Pending status as a result of the
partitions being merged.
- 5
- Inactive Pending. Cluster Resource Services is in the
process of ending on this node as the result of an End Cluster Node API call.
The node is still in the cluster membership list.
- 6
- Inactive. Cluster Resource Services has been ended on the
node as the result of an End Cluster Node API call. The node is still in the
cluster membership list, but is no longer communicating with other nodes in the
cluster.
- 7
- Failed. A previously active node has failed. A failure is
defined to be a system or clustering failure detected by Cluster Resource
Services.
- 8
- Partition. The node is only communicating with a subset of
the cluster due to a network failure detected by Cluster Resource Services
which has resulted in the loss of communications to one or more nodes in the
cluster. Once the partitioned nodes have been merged back into a whole cluster,
the node will change to Active status without operator intervention. Of course,
any node that had a status of Failed in any partition will still have a status
of Failed after the merge.