16.15.3 Returning a node to the multi-node configuration

This section explains the procedure for returning a node that was separated from a multi-node configuration, back to the multi-node configuration.

Organization of this subsection

(1) Notes on returning nodes to a multi-node configuration
(2) Procedure for returning nodes to a multi-node configuration
(3) Procedure for switching over a returned node to the master node

(1) Notes on returning nodes to a multi-node configuration

The values specified in the following server definition operands cannot be changed:
- adb_sys_rthd_num
The specified value cannot be changed even for operands that need to be specified to the same value on all nodes. For a list of operands for which the same value needs to be specified on all nodes, see 16.3.5 Creating server definitions on all nodes.

Do not change the specified values even of operands other than the preceding operands, unless absolutely necessary.
Nodes cannot be returned to a multi-node configuration if at least one of the following cases applies. Before performing an operation that returns nodes to a multi-node configuration, confirm that none of the following cases applies:
- The HADB servers that make up a multi-node configuration include ones on which termination processing is in progress.
- A master node switchover is in progress.
- There are HADB servers that are being returned to a multi-node configuration.
- There are application programs or commands that are connected to an HADB server.

To Page Top

(2) Procedure for returning nodes to a multi-node configuration

The procedure for returning nodes to a multi-node configuration is as follows:

Procedure:

Execute the HA Monitor monshow command on the node subject to the return

Make sure that monitoring by HA Monitor is stopped.

Important

Perform the following procedure after making sure that monitoring by HA Monitor is stopped. If you accidentally execute the HA Monitor monbegin command from step 4 before executing the adbstart command from step 3, execute the HA Monitor monend command (make sure to stop monitoring by HA Monitor, and any running HADB servers). Afterwards, perform the procedure for returning nodes to a multi-node configuration from the very beginning.
Execute the adbchgsrvmode --maintenance command on the master node

Change the HADB server operation mode on the master node to maintenance mode. Alternatively, make sure that there are no ongoing transactions.
Execute the adbstart command on the node subject to the return

After the adbstart command is executed, the KFAA91109-I message is output asking you to execute the HA Monitor monbegin command.
Execute the HA Monitor monbegin command on the node subject to the return

After confirming that the KFAA91109-I message has been output, execute the monbegin command from a different terminal (command input screen).

Important

Before the adbstart command terminates, execute the monbegin command.

When the adbstart command terminates normally, returning of the node to the multi-node configuration is complete.
Execute the adbchgsrvmode --normal command on the master node

If you changed the master node's HADB server operation mode to maintenance mode in step 2, return the operation mode to normal mode.
Execute the adbls -d node command on any node

Refer to the NODE_TYPE field in the command execution result, and make sure that there is one master node. If all the nodes are slave nodes, execute the HA Monitor monact command on the node you want to make the master node. The node where you executed the monact command becomes the master node.

Note

In step 2, if the master node stops at the moment you execute the monbegin command before the KFAA91109-I message is output, all nodes might become slave nodes.

Note that nodes are always returned to a multi-node configuration as slave nodes. To set the node you returned as the master node, see (3) Procedure for switching over a returned node to the master node.

Important: In environments where synonym searches are used, you need to synchronize the synonym dictionary files after returning the nodes. For details about how to synchronize the synonym dictionary files, see 16.24.5 Synchronizing synonym dictionary files.

Note: If the audit trail facility is enabled, audit trail output will automatically resume for the node when the process of returning it has completed. If a node is returned to a multi-node configuration where the audit trail facility is disabled, the node will be returned with the audit trail facility disabled even if the facility was enabled when the HADB server of the node was last operational.

To Page Top

(3) Procedure for switching over a returned node to the master node

The following explains the procedure for switching over a returned node to the master node. The procedure differs depending on whether you are using host reset or SCSI reservation for shared disk.

Procedure (when using host reset)
1. Return the node that was separated from the multi-node configuration (the former master node) back to the configuration.
  
  Note that nodes are returned to a multi-node configuration as slave nodes.
2. Execute the HA Monitor monswap command on the current master node.
  
  The current master node's HADB server terminates, and the returned node in step 1 switches over to the master node.
3. Execute the adbls -d node command on any node.
  
  Refer to the NODE_TYPE field in the command execution result, and make sure that the master node has switched over. You cannot perform the subsequent operations until the master node switchover is complete.
4. Return the node you terminated in step 2, to the multi-node configuration.
  
  For details about the procedure for returning a node to a multi-node configuration, see (2) Procedure for returning nodes to a multi-node configuration.
Procedure (when using SCSI reservation for shared disk)
1. Return the node that was separated from the multi-node configuration (the former master node) back to the configuration.
  
  Note that nodes are returned to a multi-node configuration as slave nodes.
2. Execute the adbstop --cancel --node command on the master node.
  
  The master node terminates normally.
3. Execute the HA Monitor monswap command.
  
  When you execute the preceding command on the node you terminated in step 2, the node you returned in step 1 becomes to the master node.
4. Execute the adbls -d node command on any node.
  
  Refer to the NODE_TYPE field in the command execution result, and make sure that the master node has switched over. You cannot perform the subsequent operations until the master node switchover is complete.
5. Return the node you terminated in step 2, to the multi-node configuration.
  
  For details about the procedure for returning a node to a multi-node configuration, see (2) Procedure for returning nodes to a multi-node configuration.

To Page Top