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

F.2.19 sfcproxyop(1M) Executes operation for high speed backup and restoration

[SYNOPSIS]

sfcproxyop backup [ -s ] rel_name

sfcproxyop part rel_name

sfcproxyop rejoin [ -s ] rel_name

sfcproxyop mount rel_name mount_point

sfcproxyop umount rel_name mount_point

[DESCRIPTION]

sfcproxyop executes operation for high-speed backup and restoration. Before sfcproxyop is executed, the relationship between the backup source and destination must have been set up with the sfcproxyjoin(1M) .

sfcproxyop includes the following subcommands:

• backup - Backup in OPC operation
• part - Snapshot creation by separation in EC operation
• rejoin - Rejoin of backup destination in EC operation
• mount - Mount the backup destination file system
• umount - Unmount the backup destination file system

The following explains the function of each subcommand.

[ backup ] - Backup in OPC operation

The backup subcommand executes One Point Copy of the file system by copy from the backup source to the backup destination specified by the relationship rel_name in the OPC operation. When sfcproxyop is executed, the following conditions must be satisfied.

• The file system of the backup source must be unmounted in all nodes. Or, the file system must be mounted in the node on which sfcproxyop is executed. If the file system is active and it is not mounted in the node on which sfcproxyop is executed is inactive, sfcproxyop fails.
• Real copy processing must not have been executed at the backup source or destination.
• The file system of the backup destination must be unmounted.
• The backup destination volumes of GDS must be stop status.
• The backup source volumes of GDS must not be data invalid (INVALID) status.

The copy operation is performed logically instantaneously but the file system update operation at the backup source is temporarily locked while a copy command is issued.

By default, sfcproxyop returns without waiting for the real copy processing to be completed. When the -s option is specified, sfcproxyop returns after the real copy processing is completed. Even in this case, the file system is locked only while the copy command is issued.

[ part ] - Snapshot creation by separation in EC operation

In the EC operation, part subcommand instantaneously separates the backup from the backup source specified with the relationship rel_name and creates a file system snapshot at the backup destination. When sfcproxyop is executed, the following conditions must have been satisfied.

• The file system of the backup source must be unmounted in all nodes. Or, the file system must be mounted in the node on which sfcproxyop is executed. If the file system is active and it is not mounted in the node on which sfcproxyop is executed is inactive, sfcproxyop fails.
• The backup destination must have been joined in the backup source.

Although the separation operation is performed logically instantaneously, the file system update operation at the backup destination is temporarily locked while a separation command is issued.

After sfcproxyop returns, the equivalent status with the backup source is cleared, making it possible to use the file system of the backup destination and the partitions that compose that file system. To return to the equivalent status after using the backup destination, use rejoin subcommand.

[ rejoin ] - Rejoin of backup destination in EC operation

In the EC operation, rejoin subcommand starts copy from the backup source to the backup destination specified by the relationship rel_name and returns to the join status. When sfcproxyop is executed, the following conditions must be satisfied.

• Real copy processing must not have been executed at the backup source or destination.
• The file system of the backup destination must be unmounted.
• The backup destination volumes of GDS must have stopped.
• The backup source and destination must be in separate status.
• The backup source volumes of GDS must not be data invalid (INVALID) status.

By default, sfcproxyop returns without waiting for the real copy. When the -s option is specified, sfcproxyop returns after the real copy processing is completed.

After sfcproxyop returns, the backup destination becomes join status to the backup source. The file system of the backup destination and partitions that compose that file system cannot be used until the separation operation is performed with sfcproxyop part or the relationship is canceled with sfcproxybreak(1M).

[ mount ] - Mount the backup destination file system

Mount the GFS Shared File System created with backup subcommand or part subcommand at the mount point specified by mount_point.

In the EC operation, the backup source and destination must be in separate status with part subcommand.

To mount the file system created at the backup destination, use mount subcommand all the time.

The following is different from normal file system and needs to be cared.

• The file system cannot be mounted on two or more nodes at the same time. However, it can be mounted on the node that is not defined as MDS node.

[ umount ] - Unmount the backup destination file system

Unmount the backup destination GFS Shared File System that mounted at the mount point mount_point.

[OPTION]

The following option can be specified:

-s

Synchronization specification. This option waits for the real copy processing to be completed.

[OPERAND]

The following operand can be specified:

rel_name

Specify an operation target relationship with the relationship name rel_name.

[EXAMPLES]

(Example 1)

To separate the backup destination from the backup source specified with the relationship with the relationship name backup1_for_volume0001 in the EC operation, execute the following command.

# sfcproxyop part backup1_for_volume0001

(Example 2)

To rejoin the backup destination specified with the relationship with the relationship name backup1_for_volume0001 and put it in the equivalent status in the EC operation, execute the following command.

# sfcproxyop rejoin backup1_for_volume0001

(Example 3)

To perform instant copy from the backup source to the backup destination specified with the relationship name backup1_for_volume0002 in the OPC operation, execute the following command.

# sfcproxyop backup backup1_for_volume0002

(Example 4)

To mount the backup destination specified with the relationship with the relationship name backup1_for_volume0001 at the mount point /mnt, execute the following command.

# sfcproxyop mount backup1_for_volume0001 /mnt

(Example 5)

To unmount the backup destination specified with the relationship with the relationship name backup1_for_volume0001 at the mount point /mnt, execute the following command.

# sfcproxyop umount backup1_for_volume0001 /mnt

[EXIT STATUS]

The following exit values are returned:

0

Successful completion.

>0

An error occurred.

[SEE ALSO]

sfcproxybreak(1M), sfcproxyinfo(1M), sfcproxyjoin, sfcproxyjoin(1M), sfcproxyrestore(1M)

sdxproxy(1M) in the "PRIMECLUSTER Global Disk Services Configuration and Administration Guide."

[NOTES]

• If a backup is created with sfcproxyop backup or part while some of nodes that share the target file system are stopped, mounting of the backup file system by sfcproxyop mount may fail in the nodes that were stopped when the backup was created.
  If that happens, mount the file system in the nodes that were operating when the backup was created.
• If EC operation is performed with software only and some of the shared nodes that share the file system are stopped after they are separated by sfcproxyop part, then when the nodes are joined with sfcproxyop rejoin, the differential information may be lost and a full copy my be performed. As much as possible, stop nodes only when they are joined.
• In the EC operation, if sfcproxyop part is executed during the execution of the real copy (equivalent copy) after sfcproxyjoin equivalent or sfcproxyop rejoin is executed, the execution waits until the equivalent copy is completed. Even in this case, the file system update operation at the backup source is locked only while the separation command is issued.
• Optional product PRIMECLUSTER Global Disk Services Snapshot is required for this function.