This section provides general notes on backup and replication.
The locale (software localization information) of the internal code system (the code system specified when AdvancedCopy Manager is installed on the Management Server) used by AdvancedCopy Manager must be installed on all Managed Servers.
Action is required if there are differences between the language environments (LANG) of the Management Server and the Managed Servers. The following table shows the actions required for various combinations:
Management Server | Managed Server | Action required |
---|---|---|
Windows (SJIS) | Windows (SJIS) | None |
Solaris (EUC) | Install the SJIS package on the Managed Server. | |
HP-UX (EUC) | Install the SJIS package on the Managed Server. | |
Linux (EUC) | None | |
Linux (UTF8) | None | |
AIX (SJIS) | None | |
AIX (EUC) | Add the SJIS language environment on the Managed Server. | |
In accordance with the Solaris locale setting (standard: EUC) | Windows (SJIS) | None. (If the Management Server is EUC, Managed Server processing is problem free.) |
Solaris (EUC) | None (if the code is the same). If the Management Server is SJIS, install the SJIS package on the Managed Server. | |
HP-UX (EUC) | None (if LANG is the same). If the Management Server is SJIS, install the SJIS package on the Managed Server. | |
Linux (EUC) | None | |
Linux (UTF8) | None | |
AIX (SJIS) | None (if LANG is the same). If the Management Server is EUC, add the EUC language environment. | |
AIX (EUC) | None (if LANG is the same). If the Management Server is SJIS, add the SJIS language environment. | |
Linux (EUC) | Windows (SJIS) | None. (If the Management Server is EUC, Managed Server processing is problem free.) |
Solaris (EUC) | None | |
HP-UX (EUC) | None | |
Linux (EUC) | None | |
Linux (UTF8) | None | |
AIX (SJIS) | Add the EUC language environment on the Managed Server. | |
AIX (EUC) | None | |
Linux (UTF8) | Windows (SJIS) | None |
Solaris (EUC) | Add the UTF8 language environment on the Managed Server. | |
HP-UX (EUC) | Add the UTF8 language environment on the Managed Server. | |
Linux (EUC) | None | |
Linux (UTF8) | None | |
AIX (SJIS) | Add the UTF8 language environment on the Managed Server. | |
AIX (EUC) | Add the UTF8 language environment on the Managed Server. |
After the appropriate package is installed, re-execute the [Discovery/Reload] operation for the Managed Server at the Web Console.
A transaction volume and a backup volume must have the same size.
However, not all of the size specified by the fdisk command will be usable for the following disks because the OS control space is automatically allocated.
Primary partition in the beginning of a disk (the leftmost on the fdisk command)
All the extended logical disks
Figure 9.1 Available Size of a Partition
Create partitions for the transaction and backup volumes in the same size and combination as shown below.
Figure 9.2 Combination of Transaction and Backup Volume
Figure 9.3 Example of a Disk with Partitions of Different Sizes
If you are creating the source and destination volumes in the same size for replication, create partitions in the combination as shown above.
Point
To prevent an operational error, you are recommended to define as extended logical disks all the logical disks on the ETERNUS Disk storage system to be managed by AdvancedCopy Manager.
If you cannot define all the logical disks as extended logical disks, you are recommended to use the same configuration for a partition, including a device defined as a transaction volume and a partition with a device defined as a backup volume.
Note
To backup a GDS volume, specify the sector size to set a backup volume size that matches the GDS volume.
Create the backup volume as follows:
Find out the accurate size and number of the GDS volumes to be backed up.
The accurate size is obtained by multiplying the number of blocks output by the following command by 512 bytes.
sdxinfo -o <volume-name> |
In the following example, 425984 is output as the number of blocks. Therefore, the size is 218103808 bytes (425984 blocks x 512).
# sdxinfo -o v1 OBJ NAME CLASS GROUP SKIP JRM 1STBLK LASTBLK BLOCKS STATUS ------ ------- ------- ------- ---- --- -------- -------- -------- -------- volume v1 c1 g1 off on 65536 491519 425984 ACTIVE |
Use the parted command to check that the disk format of the backup volume is msdos. If the format is not msdos, execute "/sbin/parted <device name> mklabel msdos" to format as msdos.
Find out the number of bytes in each sector of the disk where the backup volume is to be created.
In the following example, "sectors of 512 bytes" indicates that the size of each sector is 512 bytes.
# /sbin/sfdisk -uS -l /dev/sdp Disk /dev/sdp: 1017 cylinders, 133 heads, 62 sectors/track Units = sectors of 512 bytes, counting from 0 Device Boot Start End #sectors Id System /dev/sdp1 0 - 0 0 Empty /dev/sdp2 0 - 0 0 Empty /dev/sdp3 0 - 0 0 Empty /dev/sdp4 0 - 0 0 Empty |
The following error is returned if the disk does not have an msdos disk label:
"sfdisk: ERROR: sector 0 does not have an msdos signature" |
Use the fdisk command or the parted command in advance to create an msdos disk label on the targeted disk.
Example: Use the # parted command to create an msdos disk label.
# /sbin/parted /dev/sdp mklabel msdos |
Calculate the number of sectors in the backup volume being created.
218103808 bytes / 512 bytes = 425984 sectors |
However, the partition descriptor is located in the first sector of a logical volume, so this sector cannot be used. Therefore, add one sector to the number of sectors obtained in this calculation.
425984 sectors + 1 = 425985 sectors |
Thus, the number of sectors to create is 425985.
Use the sfdisk command to create the files to be input.
Since files need to be created as logical partitions in extension partitions in a backup volume, files like the following are created:
# cat ./sdp-partition.dat ,,E ; ; ; ,128,L ,425985,L ,425985,L ,425985,L |
The first logical partition (/dev/sdp5) cannot be made the same size as the other logical partitions for reasons related to partition descriptors.
Therefore, the first logical partition is created to be the appropriate size (in this case, 128 sectors) and is not used as part of the backup volume.
Use the sfdisk command to create the partitions.
# /sbin/sfdisk -uS --force /dev/sdp < ./sdp-partition.dat Checking that no-one is using this disk right now ... OK Disk /dev/sdp: 1017 cylinders, 133 heads, 62 sectors/track Old situation: Units = sectors of 512 bytes, counting from 0 Device Boot Start End #sectors Id System /dev/sdp1 0 - 0 0 Empty /dev/sdp2 0 - 0 0 Empty /dev/sdp3 0 - 0 0 Empty /dev/sdp4 0 - 0 0 Empty New situation: Units = sectors of 512 bytes, counting from 0 Device Boot Start End #sectors Id System /dev/sdp1 1 8386181 8386181 5 Extended /dev/sdp2 0 - 0 0 Empty /dev/sdp3 0 - 0 0 Empty /dev/sdp4 0 - 0 0 Empty /dev/sdp5 2 129 128 83 Linux /dev/sdp6 131 426114 425984 83 Linux /dev/sdp7 426116 852099 425984 83 Linux /dev/sdp8 852101 1278084 425984 83 Linux Warning: partition 5 does not end at a cylinder boundary Successfully wrote the new partition table Re-reading the partition table ... If you created or changed a DOS partition, /dev/foo7, say, then use dd(1) to zero the first 512 bytes: dd if=/dev/zero of=/dev/foo7 bs=512 count=1 (See fdisk(8).) |
Use the created /dev/sdp6, /dev/sdp7, and /dev/sdp8 as the backup volume.
The following devices must not be backed up or replicated:
The device on which AdvancedCopy Manager has been installed
The device on which the management list of AdvancedCopy Manager resides
Logical volume of LVM(Logical Volume Manager) or volume group
Data in transaction volume | Maintenance of data integrity | Operation |
---|---|---|
Oracle database space being used | Linkage with Oracle is required. | Use a sample script designed for linkage with Oracle. For details, refer to "Appendix B Sample Script for Backup Operations". |
File system | A AdvancedCopy Manager command unmounts the file system to maintain the integrity. | Refer to "Appendix A Pre-processing and Post-processing of Backup and Restoration" and "Appendix C Pre-processing and Post-processing of Replication". |
Other than the above | The data integrity must be maintained in operations. | Take appropriate action, such as stopping transactions during execution of backup or replication. |
Before starting either the backup or replication operation, "3.4.4 Fetching device information on a Managed Server" that is managed by all Managed Servers using the Web Console. The time taken to complete this operation is proportional to the total number of devices defined on the selected Managed Servers. If many devices are involved, perform this operation while the CPU or I/O load on the system is low.
As a guide, about 0.5 seconds are required per device (i.e., partition) when there is no load on the system.
An upper limit (maximum value) applies to the number of Advanced Copy sessions in one Logical Unit (LU) that can be executed simultaneously. Ensure the number of volumes in one LU is the same or less than the maximum number of copy sessions per LU.
When you create more volumes in one LU than the maximum number of copy sessions per LU, for the volumes in that LU, an operational configuration is required that does not implement Advanced Copy simultaneously in excess of the maximum number of copy sessions per LU.
Refer to the ETERNUS Disk storage system SA/SE Handbook for the maximum number of copy sessions per LU.
Before backup/restoration or replication of a volume that can be mounted from multiple servers, unmount and cancel any pending mount on the other servers.
For disks that need not be mounted from other servers, prevent multiple servers from detecting or accessing the same partition by setting up hardware, such as devices in the ETERNUS Disk storage system and Fibre Channel switches, accordingly.
For a copy operation targeted at a file system, unmount the relevant volume to prevent data access and maintain the data integrity.
If backup/restoration and replication are performed without unmounting the volume, an error occurs since a volume in use cannot be unmounted.
Note the following about processing where unmounting is enabled:
Another application is not using the volume. If another application is using it, suspend the application temporarily.
Users are not using the volume. If users are using it, prohibit them from using it temporarily.
Another volume is not mounted on the volume. If it is mounted, unmount it temporarily.
The volume is not shared with NFS sharing. If it is shared, unshare it temporarily.
Unmounting is required only during command execution. Normal operations can be resumed after command execution.
For box-to-box mirroring, both systems of the mirroring must be backed up if restoration is required by OPC in the event of a box failure. In these cases, the size of the backup volume must be a physical volume size, not logical volume size.
Logical Unit (disk unit) copy can be used for replication operations.
Note the following points if Logical Unit (disk unit) copy is executed:
Both servers must run Version 13.1 or later.
When copying between Logical Units (disks) of different sizes, the physical copy matches the size of the smaller Logical Unit (disk). If the copy direction is from the smaller Logical Unit (disk) to the larger Logical Unit (disk), part of the area on the larger disk is not physically copied to, and data from before the copy operation remains. If this remaining area contains the Logical Unit (disk) management information, partitioning tools may identify the disk format incorrectly. Therefore, format the copy destination Logical Unit (disk) before copying to it.
To copy the GPT disks by each Logical Unit (disk), create the source and destination volumes in the same size for replication in the unit of bytes.
When copying between volumes of different sizes, the GPT label of the destination volume may be changed. If you restore the destination volume whose GPT label is changed, you cannot use the destination volume.
If the replication source/destination volume areas for replication management, and the transaction/backup volume areas for backup management duplicate each other, replication management and backup management operations cannot be performed simultaneously for volumes that have duplicate areas.
Execute copy operations when the partitions included in the Logical Unit (disk) are in the unmounted state. If copy is executed when the partitions are in the mounted state, the partitions included in the copy destination disk may not be able to be made consistent as a file system.
Logical Unit (disk unit) copying cannot be used for SDX objects.
Copies can be made between different operating systems by using replication operations.
Use the following combinations to execute copy between different operating systems:
Copy source | Copy destination | |
---|---|---|
slices | Logical Unit (disk) | |
Partition | A | N/A |
Logical Unit (disk) | A | N/A |
(A=Available, N/A=Not Available)
Note
If SnapOPC+ is to be used for executing copy operations between different operating systems, perform copy between operating systems that can use SnapOPC+.
The MBR disk format (recommended) and GPT disk format are supported.
Note the following points if AdvancedCopy Manager operations use device names generated by the udev facility.
The device names used when a file system is mounted must be the same as the device names used by AdvancedCopy Manager.
If AdvancedCopy Manager operations use by-id, use by-id to mount the file system.
If AdvancedCopy Manager operations use by-path, use by-path to mount the file system.
If the udev name is used for replication between servers, the AdvancedCopy Manager version must be the same as on the other server.
If the other server has the following operating systems, the udev name cannot be used for replication between servers:
HP-UX
AIX
When replication occurs between servers, the -m option can be specified in the following commands in order to avoid communication processing to non-operating servers:
When the -m option is specified, any one of the following conditions must apply in order to ensure that problems do not arise as a result of not performing volume pre-processing and post-processing at the non-operating server:
The non-operating server has been shut down.
Manual pre-processing and post-processing is performed for the volumes on the non-operating server.
The target volume is a RAW device (Oracle, etc., that does not need pre-processing and post-processing), rather than a file system.
The Advanced Copy processing is performed by the ETERNUS Disk storage system. Therefore, direct enquiries concerning copy processing performance to the support department for the ETERNUS Disk storage system.
In the case where a source volume has a filesystem label and where the following conditions hold, Advanced Copy can cause the status where multiple volumes with the same label exist in one system.
Backup/restore or replication within a server is employed.
Replication between servers is employed and a volume in the destination server has the same label as a source volume.
When there are multiple volumes with the same label, processing such as mount may not work correctly.
To avoid this phenomenon, specify not a label but a device (e.g. /dev/sda1) for /etc/grub.conf and /etc/fstab.