Configure Replication

After you complete the first three Configuration Wizard steps, the first replication step is to create file systems: the blockpool file system(s), and the source and target file systems you plan to use.

Note: Although StorNext supports replicating from multiple source hosts and file systems to multiple target hosts and file systems, for simplicity this procedure describes how to replicate between one source and one target file system on the same host.

1. If you have not already done so, launch the StorNext Configuration Wizard and proceed through Welcome, License and Name Servers steps to the File System step.
2. The Configuration > File System page appears.

3. On the Configuration > File System screen, click New. The Configuration > File System > New page appears.

   

4. At the File System Name field enter the name of a file system to be used as a replication source. A default mount-point path automatically appears but you can change this mount point if you wish.
5. Choose the Replication/Deduplication option. A warning message alerts you that “A blockpool has not been created.” Disregard this message for now because you will create the blockpool file system in the Step 2: Set Up the Blockpool.

6. Select the Generate option, and then click Continue to proceed.

   

7. Select a set of LUNs for the file system, and then click Assign.

8. Click Continue.

   

9. If desired, click the arrows beside the Configuration Parameters and Stripe Group/Disk Management headings to view information.
10. Click Apply to save the new file system. For more information about creating file systems, see File Systems OverviewFile Systems Overview.

Note: If you use either the StorNext GUI or the snpolicy command to create or modify a replication/deduplication policy, a policy text file is written to the file system. For example, suppose that /stornext/photos/ is the mount point for file system named photos. If a policy named pol_replicate_1 is created in that file system, a text copy of the policy information called /stornext/photos/.rep_private/config/ pol_replicate_1 is created. If the file system is damaged and has to be recreated, the policy must also be recreated. This is simpler to do beginning with the StorNext 4.1 release because a backup copy of the policy text file is saved whenever a policy is created or updated. (The backup copy is saved as a file named /usr/cvfs/data/fsname/policy_history/policyname.date_time) In the previous example, the file system name (fsname) is photos and the policy name is pol_replicate_1. So, the backup copy would have a name such as /usr/cvfs/data/photos/policy_history/pol_replicate_1.2010-10-29_14-07-13. The backup copy directory is not in the same file system as photos. If Storage Manager is used on the machine, all the policy backup files will be backed up along with the regular Storage Manager backups. Quantum recommends executing the command snpolicy_gather –b > some_file after upgrading to the latest version of StorNext. This saves a copy of your current configuration. The –b option creates a copy of policy information in the usr/cvfs/data/fsname/policy_history directory.

Create a Target File System and Blockpool File System

1. Repeat the above file system creation process (sub-Steps 1 - 8) and create the file system you intend to use as a target for replication on this same server.

   Note: The Replication/Deduplication option (in sub-Step 5) must be enabled for both source file system and target file system. If a file system (source and/or target) is also to be used for Storage Manager, Data/Migration option (in sub-Step 5) must be enabled.

2. Configure another file system for the Blockpool that has neither Data Migration nor Replication/Deduplication enabled.

   Note: The step above assumes that the source file system and target file system reside on the same machine. If the target file system resides on a different machine (target machine), follow the above file system creation process (sub-Steps 1 - 8) on the target machine where the target file system for replication resides. In addition, configure another file system for the Blockpool on the target machine that has neither Data Migration nor Replication/Deduplication enabled.

In this step, set up the blockpool on the blockpool file system you just created in the previous step. If the source file system and target file system reside on different machines, perform the following sub steps on both source machine and target machine.

1. Choose the StorNext Configuration Wizard’s Storage Destinations task. The Configuration > Storage Destinations page appears. There are four tabs on this screen: Library, Storage Disk, Replication Targets, and Deduplication. When configuring replication we are concerned with the Replication Targets and Deduplication tabs. (The deduplication infrastructure is used to handle the transfer of file data for the replication feature, so it must be configured even when the deduplication feature is not used.)

2. Click the Deduplication tab. The Configuration > Storage Destinations > Deduplication page appears.

   

3. Click the Deduplication tab. This tab has only one field called Blockpool Host File System. At this field select from the dropdown list the file system to use for the blockpool (this is the file system you created in the previous step).

   Note: Once applied, the blockpool location cannot be moved to another file system. Be certain of the blockpool location before you continue.

4. After you select the blockpool file system, click Apply. A background job is started to create the blockpool.

In this step you will specify the actual targets to which you want replicated data sent. Namespace realization will also occur on these targets.

1. Click the Replication Targets tab. The Configuration > Storage Destinations > Replication Targets page appears.

2. Click Add Host.

   

3. At the Hostname or IP field, enter the host name or its IP address. If the target is an HA cluster, the address should be the vIP for that cluster. If multiple vIPs are configured for the target HA cluster, select one vIP address that is accessible from the source host.
4. Click Scan Host to populate the Mount Point box with appropriate file systems which are configured for replication/deduplication.
5. Select the file system you created for use as the target in Step 1: Create Source and Target File Systems, and then click Add.
6. Click Apply. At this point you should see your file system listed as a replication target.

Note: If you were adding additional replication targets, you would repeat steps 3 - 6 to add additional hosts and file systems.

(Optional) Configuring Replication for an HA System

• If you are planning to use replication on a high availability (HA) system, this is the point in the configuration process when you should configure HA. If you do not configure HA here, misconfiguration could result and you could be prevented from using replication on your HA system.
• If you are using replication on an HA system, proceed to Configure Multilink, and then return to Step 4: Create a Replication Storage Policy.
• If you are not using replication on an HA system, proceed to Step 4: Create a Replication Storage Policy.

The next step in configuring replication is to create a replication storage policy. This policy contains the replication "rules" specific to your replication source and target file systems. You must create a replication policy for the source directory and enable inbound replication for the target file system.

See Add a Replication or Deduplication Policy.

The preceding four configuration steps accomplished the following:

• Created a source replication policy and associated a source directory with it
• Selected a target file system on a target host machine and left the target directory unspecified, which uses the directory name of the source
• Set a replication schedule that runs every day at midnight
• Enabled inbound in the target policy
• Enabled outbound replication in the source policy

The contents of the source directory (additions and deletions) will now be replicated to the target directory every night. You can test this by running the policy manually at any time on the Configuration > Storage Policies screen. Select the policy you want to test and then click Run.

The Replication Blackout feature provides bandwidth management by allowing you to select of a time period when you do not want replication to run. When a blackout is not in effect, replication data transfer occurs automatically in the background as data changes in the source directories, but the replicated files do not appear in the target directory until the replication policy is run.

You can set a blackout period on the source or target file system (or both) in the file systems’ global policy. During the blackout period, both replication data transfer and the realization of file copies on the target are prevented from starting.

A blackout period for a source file system prevents automatic starting new data transfers or scheduled policies. However, manually started policies do run, and perform the necessary data transfers.

A blackout period for a target file system prevents all inbound data transfers from starting, which blocks both manually and automatically started source policies.

However, note the following caveats about blackouts:

• Any replication attempt (whether scheduled or initiated from the command line) which starts during the blackout on the source will not be started unless the force option is used. Replications started before the blackout should complete.
• Any replication request which arrives at the target during its blackout will be rejected by the target. The source will retry replication until the process succeeds. Replications started before the blackout will complete.

Configure a Replication Blackout Window

1. Choose the StorNext Configuration Wizard’s Storage Policies task.
2. On the Storage Policies screen, select the “global” policy for the desired source or target file system, and then click Edit. The Configuration > Storage Policies > Edit page appears.
3. Click the Blackout tab.
4. Click the box to the right of the Replication Blackout Window heading to display scheduling fields.
5. Specify the weekday(s), month(s), day(s), hour(s) and minute(s) when you would like to block replication from starting automatically.
6. Click Apply to save the changes in the replication/deduplication storage policy.

When the High Availability (HA) feature is used with replication, a virtual IP (vIP) address must be configured to allow a replication source to use a single IP address to access whichever server is currently performing the Primary function in the target HA cluster.

The vIP is automatically moved to the correct server as part of the failover process for the HaShared file system. See Virtual IP (vIP) for an expanded definition.

It is easiest to set up the vIP during the initial HA conversion. The vIP configuration items appear automatically at this time if a license exists for replication. It is not necessary to have a replication policy configured.

The IP address used for the vIP must be statically allocated and routable to the physical network interface of both servers in the HA cluster. Please request this IP address and netmask information from your network administrator before starting the HA conversion.

Note: This step describes only the tasks necessary for configuring replication on an HA system. For general instructions about configuring HA, see Converting to HA.

1. Choose High Availability > Convert from the Tools menu. The HA (Convert) page appears.

   

2. At the Shared File System field, select from the dropdown list the file system that will be dedicated to StorNext HA internal functions.
3. At the MDC Address field, select from the dropdown list the primary system’s IP address for use in communicating between HA MDCs.
4. Since this HA system runs a blockpool, you must configure a Virtual IP Address (vIP). Under the heading Virtual Network IP Address Configuration, check Enable and then enter the vIP (virtual IP) Address and vIP Netmask provided by your network administrator.
5. Click Convert to convert the primary node to HA.
6. When the confirmation message appears, click Yes to proceed or No to exit without converting.
7. When a message informs you that the operation was completed successfully, click OK. The configuration items for the Secondary System will be added to the page.
8. At the System Name field, enter the IP address of the Secondary System to use for communications between HA MDCs, and then click Scan Host.
9. Select the IP address of the physical interface to associate with the vIP, and then click Convert.
10. When the confirmation message appears, click Yes to proceed or No to exit without converting.
11. When a message informs you that the conversion was completed successfully, click OK to continue.

The default location of the blockpool server process is localhost. This is not sufficient for HA Clusters where the blockpool server moves with the Primary status to the redundant server in a failover of the HA Shared file system.

1. Return to the StorNext Configuration Wizard’s Storage Policies task.
2. Locate the Deduplication/Replication file system, select its global policy, and then click Edit. This step must be repeated for each Deduplication/Replication enabled file system.
3. Click the Deduplication tab.
4. At the Address for Replication and Deduplication field, click the Inherit button.
5. Replace the localhost value with the vIP address in the Override box (Override appears after you click Inherit). If multiple vIPs are configured, select one vIP address for the blockpool server.
6. Click Apply.
7. When the confirmation message appears, click Yes to proceed or No to exit. In this case you can safely ignore the warning about associated directories.
8. When a message informs you that the operation was completed successfully, click OK to continue.
9. Repeat steps 2 thru 8 for each file system.

Note: The IP address of the Blockpool Server in HA Clusters can also be configured from the tab Outbound Replication and Inbound Replication for the "global" policy. There is no difference in configuring from different tabs and the setting of the IP address for the Blockpool from one tab is automatically reflected in the corresponding field of the other two tabs.

Virtual IPs are also used in an HA environment if the multilink feature is configured. A virtual IP address is required for each NIC card you use for replication.

1. Choose Replication/Deduplication > Replication Bandwidth from the Tools menu. The Tools > Replication > Bandwidth page appears.

   

2. The Replication Bandwidth screen displays a list of NIC cards available for replication. Select Enable for each NIC card you want to include in the replication process.
3. Enter the following fields:
   • VIP: Enter the virtual IP address for the NIC. (Ask your network administrator for this address as well as the virtual netmask.)
   • VIP Netmask: Enter the virtual netmask for the NIC
   • Receive Rate: Enter the maximum data reception rate (expressed in bits per second) for the replication target. When replication data is received on the target, it will not exceed this speed. For more information, see Bandwidth Throttling.
   • Transmit Rate: Enter the maximum data transmission rate (expressed in bits per second) for the replication source. When replication data is transmitted to the target it will not exceed this speed. For more information, see Bandwidth Throttling.
   • Channels: Enter the number of channels you want enabled on the NIC.
4. Click Apply to save your changes.