PRIMECLUSTER Global File Services Configuration and Administration Guide 4.2 (Solaris(TM) Operating Environment)
Contents Index PreviousNext

Part 3 Global File Services Shared File System> Chapter 24 File System Backing-up and Restoring> 24.7 High speed backup and restoration of a GFS Shared File System

24.7.8 Operation at execution time

24.7.8.1 Backup by the instant copy

In EC employment, uses the sfcproxyop part command to separate the backup destination and create a snapshot when the backup destination is joined with the backup source.

Execute the command in the nodes that share the backup source file system.

For details on sfcproxyop(1M), see Appendix F.2.19, "sfcproxyop(1M)."

1. Checking the copy status

Use the sfcproxyinfo command to check whether the backup source and destination are joined and that the copy process has been completed. Execute the command in a node that shares the backup source file system.

# sfcproxyinfo backup1_for_volume0001
relation name: backup1_for_volume0001
type: EC
status: JOIN
master                proxy                  m_status       p_status
/dev/sfdsk/class0001/rdsk/volume0001 /dev/sfdsk/class0001/rdsk/volume0001- DONE DONE

The copy process is completed if the status in the command output is JOIN and the p_status value for all volumes is DONE rather than COPY. The sfcproxyop command waits for copy processing to end before executing the separation.

2. The following example shows the specification for backing up data from the master of the file system whose relationship name is backup1_for_volume0001 to the proxy by the equivalent method. Execute this command in the nodes that share the backup source file system.

# sfcproxyop backup backup1_for_volume0001

After the sfcproxyop part command has been executed, offline backups of the backup destination file system can be created using commands such as sfcdump(1M), dd(1M), and sfcmkbkup(1M). Online backup using commands such as tar(1) and sfccpio(1) can also be created using the sfcproxyop mount command.

24.7.8.2 Separation of a backup place

EC uses the sfcproxyop part command to create the backup copy.

In EC employment the sfcproxyop rejoin command is used to reestablish the relationship between the backup source and a separated backup destination when the backup destination is separated. In the background, GDS or the hardware starts equivalence recovery of the partitions being used by the file system. An I/O load is placed on the backup source disk unit until the partitions reach equivalent status.

For details on sfcproxyop(1M), see Appendix F.2.19, "sfcproxyop(1M)."

The following example shows the specification for rejoining a file system whose relationship name is backup1_for_volume0001. Execute the command in a node that shares the backup source file system.

# sfcproxyop part backup1_for_volume0001

If the hardware copy function is not being used and even one node that shares the file system is stopped, a full data copy is performed during the equivalence recovery executed after the rejoin. If a planned node stop is to be executed, execute the operation while the backup source and destination are joined.

24.7.8.3 Rejoining the backup destination to the backup source

In OPC implementation the sfcproxyop backup command is used to execute a backup by instant copy. The hardware starts copying of the partition used by the file system in the background.

For details on sfcproxyop(1M), see Appendix F.2.19, "sfcproxyop(1M)."

The following example shows the specification for backing up the contents from the master of the file system whose relationship name is backup1_for_volume0002 to the proxy by instant copy.

1. Checking the copy status

Use the sfcproxyinfo command to check whether the copy process is completed in the backup source and destination.

# sfcproxyinfo backup1_for_volume0002
relation name: backup1_for_volume0002
type: OPC
status: SEPARATE
master                proxy                  m_status       p_status
/dev/sfdsk/class0002/rdsk/volume0002 /dev/sfdsk/class0002/rdsk/volume0002- DONE DONE

The copy process is completed if the p_status field of the command output shows DONE instead of COPY.

2. To back up the file system by instant copy, execute the following command in the nodes that share the backup source file system.

# sfcproxyop rejoin backup1_for_volume0002

After the sfcproxyop backup command has been executed, offline backup of the backup destination file system can be created using commands such as sfcdump(1M), dd(1M), and sfcmkbkup(1M). Online backup using commands such as tar(1) and sfccpio(1) can also be created using the sfcproxyop mount command. You can minimize the effects on the I/O performance of the backup source by setting up the operation so that the data of the backup destination is accessed after the copy operation by OPC is completed.

24.7.8.4 Mount of the backup destination

Use the sfcproxyop mount command to mount the backup destination created by sfcproxyop backup in OPC implementation or the backup destination created by sfcproxyop part in EC employment.

For details on sfcproxyop(1M), see Appendix F.2.19, "sfcproxyop(1M)."

The following example shows the specification for mounting the backup destination of the file system whose relationship name is backup1_for_volume0001 to /mnt. Execute this command in the nodes that share the backup source file system.

# sfcproxyop mount backup1_for_volume0001 /mnt

24.7.8.5 Unmount of the backup destination

Use the sfcproxyop umount command to unmount a file system mounted with the sfcproxyop mount command.

For details on sfcproxyop(1M), see Appendix F.2.19, "sfcproxyop(1M)."

The following example shows the specification for unmounting the backup destination of a file system whose relationship name is backup1_for_volume0001. Execute this command in nodes in which the file system was mounted.

# sfcproxyop umount backup1_for_volume0001 /mnt


Contents Index PreviousNext

All Rights Reserved, Copyright(C) FUJITSU LIMITED 2004