This section describes how to determine the cause of a failure in accessing a repository database and the actions to perform when this occurs.
Execute the steps below until the problem is resolved and no further steps are required.
Perform this procedure on the Management Server.
Check whether the RDB daemons are still running.
Execute the following command to check for any process related to the RDB daemons.
# ps -ef | grep RDBSWSTF | grep -v grep |
If no process related to the three RDB daemons exists, retry processing after restarting (start after stopping) the daemons. For an explanation on restarting the daemons, refer to "8.5.3 Starting and stopping the RDB daemon".
Check that there is sufficient free space as required to update a repository database.
Execute the following command to check the usage ratio of space required to update a repository.
# /opt/swstorage/bin/stgdbloginf |
If the usage ratio is 100%, save the database space according to "7.1.1.3.1 Saving a database". (Consequently, there should be sufficient free space required to update a repository.)
After using "7.1.1.3.1 Saving a database", restart (start after stopping) the RMI daemon. For an explanation on restarting the daemon, refer to "8.5.2 Starting and stopping the RMI daemon".
Then, perform this step again.
Check whether the "7.1.1.3.1 Saving a database" process is in progress.
Execute the following command to check for any process related to "7.1.1.3.1 Saving a database".
# ps -ef | grep stgdbdmp | grep -v grep |
If any process related to "7.1.1.3.1 Saving a database" exists, retry processing after waiting for the end of "7.1.1.3.1 Saving a database".
Check whether the capacity of repository is insufficient.
In the log file a /var/opt/FJSVswstf/log/RDBSWSTF.log file (in cluster operation, they are /var/opt/FJSVswstf/<logical-node-name>/log/RDBSWSTF.log), search for the strings "JYP5019E" or "JYP5045E" in any message taking the form "rdb: ERROR: qdgXXXXXX-"
When a character sequence exists, please extend repository space with reference to "7.2.1 Troubleshooting: Insufficient free space in a repository".
Then, Retry processing.
Check whether an I/O error or any other error has occurred in the database space.
Check the /var/opt/FJSVswstf/log/RDBSWSTF.log file, and beginning from the last line, search for any message with "rdb: ERROR: qdgXXXXX-" in its contents (also check the messages displayed at the same time). Then, search for a string taking the form of "qdgXXXXX," and check for it in "Relationships between the qdg message and recovery mode".
If "qdgXXXXX" is in the table, examine the target according to the error description corresponding to the qdg messages in "Relationships between the qdg message and recovery mode", and determine if there is a problem. If a problem is found, take corrective action as appropriate to solve the problem, and specify the appropriate settings. Then, recover the database space by using "7.1.1.3.3 Recovering a database".
After the database recovery command ends normally, acquire the latest save data by using "7.1.1.3.1 Saving a database".
Then, restart (start after stopping) the AdvancedCopy Manager daemons. For an explanation on restarting the daemons, refer to "8.5.4 Starting and stopping AdvancedCopy Manager daemons".
Retry processing.
Note
If the database recovery command is executed with the value "0" specified for -m option, perform the following tasks after the command ends normally:
Perform the resource consistency command specified with -x option (refer to swstsrsemtch (Resource match command)) on all Managed Servers in a backup operation.
Perform the resources adjustment command (refer to swsrprecoverres (Resource adjustment command)) on the Managed Servers that are performing replication using the resources adjustment command.
If an error such as an I/O error occurs in a database space, a file required for analyzing the cause of the error may be created in the <DB file directory>/SWSTFDB/core. If the corresponding recovery ends normally, delete the file in this directory.
Collect the files in /var/opt/FJSVswstf/log, and contact a Fujitsu system engineer.
