Verify with Sybase Technical Support that a restore operation is necessary. If you cannot open your database on a secondary server, synchronize the server, following the instructions in “Synchronizing secondary servers”. It is never necessary to restore a coordinator node due to secondary node problems.
You can perform read-write restore operations only against a server running the utility database. You cannot run restore operations against a secondary server.
Restoring IQ store
data when restoring database in same location
Confirm that database home directories for each server still exist. If not, create them or restore them from file system backups.
Shut down every server in the multiplex (coordinator and all secondary servers). In Sybase Central, right-click the multiplex icon and use the Stop Server wizard. The wizard lets you select servers to stop.
If automatic startup is enabled in your ODBC configuration,
users on the same machine as the server may be set up to start the
server automatically. Prevent this from happening while you are
restoring the database.
After stopping all servers, confirm that the database shut down successfully. If you see an active iqsrv15 process with name of a multiplex, stop the process.
To confirm database shutdown on a UNIX system, use the ps command. For example:
% ps -ef | grep iqsrv15
fiona 434 1 1 May 19 ?? 0:05 start_iq -n myhost_myserver -c 32MB -x tcpip(port=1234) mpxdb.db fiona 4751 442 1 16:42:14 pts/5 0:00 grep start_iq
To confirm database shutdown on a Windows system, use Task Manager. Look on the Processes tab for iqsrv15.exe or find the IQ Server icon in the system tray and stop it using right-click and Shutdown.
Move files required for debugging and reconfiguring the multiplex.
Make a file system copy of the .iqmsg file. If you have message log archiving configured, see “Backing up the right files,” in System Administration Guide: Volume 1.
On each server, preserve any files that were added to IQ_SYSTEM_TEMP for that server. These files are of the form dbname.iqtmp if you used an OS file, or they may be raw devices. If the IQ temporary store is damaged, start the server with the -iqnotemp switch to drop and recreate the temporary store dbspaces. For more information, see the Release Bulletin.
Either drop the database or delete the following files from the coordinator:
<database_home>/<dbname>.db
<database_home>/<dbname>.log
If a query server is damaged, however, drop it and re-create it after RESTORE. Then follow the instructions in “Restoring IQ store data when restoring database in a different location.”
Start the utility database from the coordinator server directory using the coordinator server’s name:
% start_iq -n thoreau_Server01 -c 32MB -x tcpip(port=1234)
Connect to the utility database (utility_db).
% dbisql -c "eng=thoreau_Server01;uid=DBA;pwd=SQL; dbn=utility_db"
Run the RESTORE command. To restore certain dbspace files to a different path, specify a RENAME clause. For details, see the RESTORE statement in Reference: Statements and Options.
Shut down the utility database.
Make sure that the temporary dbspaces exist as before, on raw devices or as files of the correct length. See “Backing up the right files,” in Chapter 12, “Data Backup, Recovery, and Archiving,” in System Administration Guide: Volume 1. For information on starting the server without using the IQ temporary store, see the Release Bulletin for your platform.
Start the coordinator server and, if restoring to the same location, synchronize the secondary servers. For more information, see “Synchronizing secondary servers”.
Start the secondary servers.
Restoring IQ store
data when restoring database in different location
Confirm that database home directories for each server still exist. If not, create them or restore them from file system backups.
If this is not the first time you have restored to the new location, shut down all multiplex servers running at the destination location (coordinator and secondary servers). In Sybase Central, right-click the multiplex and use the Stop Server wizard. The wizard lets you stop one or more servers. The multiplex at the original location where the backup was taken may continue running.
If automatic startup is enabled in your ODBC configuration,
users on the same machine as the server may be set up to automatically start
the server. Prevent this from happening while you are restoring
the database.
After stopping all servers, confirm that the database shut down successfully. If you see an active iqsrv15 process with the name of a multiplex server, stop the process.
To verify on a UNIX system, use the ps command. For example:
% ps -ef | grep iqsrv15
fiona 434 1 1 May 19 ?? 0:05 start_iq -n myhost_myserver -c 32MB -x tcpip(port=1234) mpxdb.db fiona 4751 442 1 16:42:14 pts/5 0:00 grep start_iq
To verify on a Windows system, use Task Manager. Look on the Processes tab for iqsrv15.exe or find the IQ Server icon in the system tray and stop it using right-click and Shutdown.
Move files required for debugging and reconfiguring the multiplex.
Make a file system copy of the .iqmsg file. If you have message log archiving configured, see “Backing up the right files,” in System Administration Guide: Volume 1.
Start the utility database from the coordinator server directory using the new coordinator server’s name:
% start_iq -n thoreau_Server01 -c 32MB -x tcpip(port=1234)
Connect to the utility database (utility_db).
% dbisql -c "eng=thoreau_Server01;uid=DBA;pwd=SQL; dbn=utility_db"
Run the RESTORE command with a new location path for the database. To restore certain dbspace files to a different path, specify a RENAME clause. For details, see the RESTORE statement in Reference: Statements and Options.
Start the restored database either by reconnecting to the utility_db server and specifying the restored database file name or by stopping the server and restarting it with the restored database. If you restart the server, use the single-node and override flags (-iqmpx_sn 1 -iqmpx_ov 1).
Use DROP MULTIPLEX SERVER statement to
drop all the secondary nodes. For example, DROP
MULTIPLEX SERVER node_w3_skm.
Once you drop the last secondary node, the coordinator shuts down automatically, signifying conversion to simplex.
Restart the coordinator without the single node or override switch.
Recreate all the secondary nodes with the correct location path. For example:
CREATE MULTIPLEX SERVER node_r2_skm DATABASE '/sunx3005/mpx_simdb.db' HOST 'localhost' PORT 8998 ROLE READER STATUS INCLUDED
After you create the first secondary node, the server automatically shuts down, signifying conversion to multiplex.
In the preceding command, you must include the database
file extension .DB.
When you restart the coordinator, you see a warning in the server log about the multiplex environment being invalid. This warning is generated if IQ_SYSTEM_TEMP dbspace does not contain any files, and is the case for all the secondary nodes you created in step 12. Ignore this warning for now.
Synchronize the secondary servers, following the instructions in “Synchronizing secondary servers”.
Start the secondary servers.
Connect to each secondary server and add files in IQ_SYSTEM_TEMP.
Run sp_iqmpxvalidate on
the coordinator. It should report no error detected.
To restore an exact copy of the multiplex to a different location, when copies of all of the server's temporary files exist at the new location, you replace steps 9 through 17 with:
Use ALTER MULTIPLEX SERVER to alter the server name, host, port, and database path of each server.
