2.5.7 If a problem occurs in 1-to-1 node switching systems
This subsection describes the actions that you must take with for each OS, when the node switching process for a standby host is timed out due to a failure in the database server (such as server is down or deadlock).
- Organization of this subsection
(1) In Windows
Manually keep the standby host online, after acquiring the log.
(a) Acquiring the log of 1-to-1 node switching system
If a trouble occurs in the 1-to-1 node switching system, you must acquire the cluster log. The following is the output destination of the cluster log when Windows is installed in the standard path:
C:\WINDOWS\cluster\cluster.log
The following information is output to a cluster log:
-
Operation log of cluster service
-
Error message when there is an error in the syntax of VBScript
-
Resource.LogInformation method of multi-purpose script
-
Other messages
For details on the cluster log, see the documentation on Windows.
If the J2EE server does not run, see Cosminexus Component Container log too.
(b) Manual recovery of 1-to-1 node switching system
You can manually recover the 1-to-1 node switching system according to the following procedure:
-
Cancel the cause of a timeout by restarting the database.
-
Create online target resources of the standby host.
(2) In UNIX
Resolve the cause of the timeout by restarting the database.