12.2 Preparing Snapshots for Block-Level Transfer (Linux)

We recommend that you prepare snapshots for block-level data transfer. Ensure that each volume group has sufficient free space for snapshots (at least 10% of the sum of all partitions). If snapshots are not available, Protect locks and releases each block in turn on the source workload for data transfer.

12.2.1 Configuring LVM Snapshots for Linux Volume Replication

The blkwatch driver leverages LVM snapshots if they are available. Copying blocks from the snapshot helps avoid potential open file conflicts.

For LVM storage, see Knowledgebase Article 7005872.

12.2.2 Configuring NSS Snapshots for NSS Pool Replication

For Linux workloads running Open Enterprise Server, the LVM snapshot solution is not available for NSS pools. During replication for NSS pools, Protect locks and releases each block in turn for data transfer. To avoid potential open file conflicts and to improve replication performance, you can leverage NSS pool snapshots for replication.

You can add a single unformatted disk to use for all NSS pool snapshots, or you can add a separate unformatted disk for each NSS pool. The best performance occurs when you add a separate disk for each pool. Add the disk before you set up the workload protection. You will prepare the disk to use, and PlateSpin will configure the NSS snapshots for the pool during replication.

NOTE:By default, PlateSpin uses the NLVM managed disk that has the largest amount of free space (unpartitioned space) for NSS pool snapshots. If you see the NSS pool snapshots for replication being located on the same disk as your root file system or on another disk that will be under constant disk IO, you should use the /etc/platespin/platespin.conf file to direct the NSS snapshots to an appropriate disk.

For information about how NSS snapshots work on Open Enterprise Server, see Guidelines for Using and Managing Pool Snapshots in the NSS File System Administration Guide for Linux.

To set up one or multiple disks to use for snapshots of NSS pools:

On the OES source workload, add an unformatted Linux disk to use for snapshots of all NSS pools. You can alternatively create a separate disk for each NSS pool.

The size of the disk should be about 20% of the amount of used data on the NSS pool. Adjust the size according to the amount of data change or growth that might occur during the interval of time for a replication.
For each disk that you created in Step 1, initialize the disk to be managed by NLVM.

You can use NSSMU or NLVM commands to initialize the disk. The device format can be either GPT or DOS.
- To use NSSMU:
  1. Launch NSSMU, then select Devices.
  2. Select the new disk, then press F3 to initialize it.
- To use NLVM commands:
  1. On the command line, enter
    NLVM init <device_name> [format]
You might need to specify which disk to use for each NSS pool’s snapshots. Create a platespin.conf file on the OES source workload, and associate the NSS pools with the new disks:
1. In a text editor, create a file at /etc/platespin/platespin.conf.
2. For each NSS pool, add device and size information under the Customlocation parameter using the following syntax:
```
[Customlocation]
/dev/pool/<yourPoolName>=<device>:<maxUnpartitionSize-in-MB>
```
  For example, specify the following entry for a pool named NSSPOOL to add snapshots on device sdc with a maximum size of 12228 MB.
```
[Customlocation]
/dev/pool/NSSPOOL=sdc:12288
```
Save the file.
Continue with setting up workload protection for the source OES workload.