16.16.1 Problems related to startup or termination of the HADB servers in the multi-node configuration
This subsection explains the steps to take when a problem related to startup or termination of the HADB servers occurs in the multi-node configuration.
- Organization of this subsection
(1) When the HADB servers in the multi-node configuration cannot be started normally
Determine the cause by following the steps indicated in the output message. Then, start the HADB servers in the multi-node configuration normally.
If you cannot determine the cause immediately, remove the failed node from the multi-node configuration, and then normally start the HADB servers. To remove the failed node, delete the specification about it from the adb_sys_multi_node_info operand in the server definition of the HADB servers on all nodes.
If only one node remains on which a failure did not occur, change the adb_multi_node_info operand in the server definition for the HADB server on that node into a comment. Then, start the HADB server as a system that does not use the multi-node function. In this case, make sure that the system directory is mounted to $DBDIR/ADBSYS. If the directory is not mounted, execute the OS's mount command to mount the system directory to $DBDIR/ADBSYS.
- Note
-
Adding a hash mark (#) to the beginning of the operand specification line changes the specification on the entire line into a comment.
In addition, if the node in which the failure occurred is the master node, change the master node by using either of the following methods:
-
Change the master node by changing the HA Monitor settings.
-
When starting the HADB server in a multi-node configuration, execute the monact command to change one of the slave nodes into the master node.
If there is a substitute machine, set up the substitute machine's environment so that it is the same as the machine of the node in which the failure occurred, and then start the HADB servers in the multi-node configuration.
- ▪ If the KFAA41205-E message is output
-
If the KFAA41205-E message is output containing the following information, opening of the DB area file fails:
-
The name of the symbolic link to the DB area file is displayed as the file name
-
open is displayed as the name of the system call that resulted in an error
-
ENOENT is displayed as the error number
In this case, when you added a node, the adbinit command might have been executed without applying the configuration change information of the DB area to the initialization options. Therefore, you need to re-add the node. For details about how to add nodes, see 16.17.1 Adding nodes. Pay particular attention to the explanation in 16.17.2 Notes on executing the adbinit command when adding nodes.
- Important
-
When the adbinit command is executed on the node to be added, the initialization options used on other nodes might not be usable as is. If you have changed the DB area configuration, you need to apply those changes to the initialization options.
-
(2) When the HADB servers in the multi-node configuration cannot be restarted
Determine the cause by following the steps indicated in the output message. Then, restart the HADB servers in the multi-node configuration.
If you cannot determine the cause immediately, remove the failed node from the multi-node configuration and then restart the HADB servers. To remove the failed node from the multi-node configuration, delete the specification about it from the adb_sys_multi_node_info operand in the server definition of the HADB servers on all nodes.
In addition, if the node in which the failure occurred is the master node, change the master node by using either of the following methods:
-
Change the master node by changing the HA Monitor settings
-
When starting the HADB server in a multi-node configuration, execute the monact command to change one of the slave nodes into the master node.
- ▪ If the KFAA41205-E message is output
-
For details about the action to take when the KFAA41205-E message is output, see ▪ If the KFAA41205-E message is output in (1) When the HADB servers in the multi-node configuration cannot be started normally.
(3) When an HADB server in the multi-node configuration cannot be terminated
While an application or command is connected to an HADB server in the multi-node configuration, that HADB server cannot be terminated. Wait for the application or command to finish. If there is a need to terminate the HADB servers right away, forcibly terminate the HADB servers in the multi-node configuration. For details about how to forcibly terminate the HADB server in a multi-node configuration, see (b) Forcibly terminating HADB servers in the multi-node configuration in (1) Termination procedures for HADB servers in a multi-node configuration under 16.4.2 Terminating HADB servers in the multi-node configuration.
If the HADB servers in a multi-node configuration cannot be terminated even when there is no connected application program or command, perform the following.
Procedure:
-
Execute the adbstop --force command on the master node.
When you forcibly terminate the master node's HADB server, the master node is switched over.
-
Terminate the HADB servers in the multi-node configuration normally.
For details about how to terminate the HADB server normally in a multi-node configuration, see (a) Normally terminating HADB servers in the multi-node configuration in (1) Termination procedures for HADB servers in a multi-node configuration under 16.4.2 Terminating HADB servers in the multi-node configuration.
If the HADB servers in a multi-node configuration cannot be terminated even with the above procedure, execute the adbstop --force command on all nodes to forcibly terminate the HADB servers in the multi-node configuration.