Replication Director: A SnapVault copy of a SnapMirror source (SM->SV) can result in data loss unless the filer that will contain the mirror copy is configured correctly.

Article:TECH198753  |  Created: 2012-10-22  |  Updated: 2013-05-22  |  Article URL http://www.symantec.com/docs/TECH198753
Article Type
Technical Solution


Issue



NetApp filers require that a specific Data ONTAP setting be made in order to successfully cascade replications from SnapMirror to SnapVault. (SM->SV)
The configuration setting ensures that the replicated data is correct and identical in the SnapVault copy.
Failure to set this value can result in data loss in a SM->SV configuration only.


Error



If this setting is not configured, the SnapVault copy will be created to match the data on the primary at the time that the SnapVault replication is run instead of replicating the contents of the mirror copy.
No status code or error indicates that a problem has occurred.


Environment



This problem is seen only in a Replication Director 7.5 environment that uses NetApp filers and where a SnapVault copy is made from a SnapMirror source (SM->SV).


Cause



The following scenario describes the problem:   

Time 0:

Take snapshot on primary.

Time 1:

Start Replication job (SnapMirror).

Time 2:

End Replication job (SnapMirror).

Time 3:

Modify data on primary.

Time 4:

Start Replication job (SnapVault created from SnapMirror source).

The problem arises at this point: The SnapVault will contain data from the primary at Time 4, and not data from the mirror copy itself, which represents the primary at Time 0.
This situation may sometimes cause Backup From Snapshot jobs to fail (see Example 2). 

 


Solution



Log in to each filer that will contain the mirror copy for the SM->SV storage lifecycle policy.
Run the following command on each those filers:

options snapvault.snapshot_for_dr_backup named_snapshot_only

See the ramifications of the problem in the following examples.
 

Example 1:  Potential for silent data loss
The backup selection contains the following directory and file:
     /mnt1/A
     /mnt1/B/file2.txt

The SLP contains the following hierarchical operations: Snapshot, Replication, Backup From Snapshot.
 

Time 0:

Take snapshot on primary.

/mnt1/A
/mnt1/B/file2.txt

Time 1:

Start Replication job (SnapMirror).

Time 2:

End Replication job (SnapMirror).

Time 3:

Modify data on primary.

The following file is removed:
/mnt1/A/file1

Time 4:

Start Replication job (SnapVault created from SnapMirror source).

SnapVault incorrectly makes the copy from the SnapMirror source, and not the SnapMirror itself. Since the source no longer contains /mnt1/A/file1, the SnapVault copy does not contain /mnt1/A/file1.

The Backup From Snapshot operation will not contain /mnt1/A/file1 either and the backup job (from the Backup From Snapshot operation) will not fail.

 

Example 2:  Potential for data loss, but status 71 received
The same backup selection and SLP setup as in Example 1.

Time 0:

Take snapshot on primary.

/mnt1/A
/mnt1/B/file2.txt

Time 1:

Start Replication job (SnapMirror).

Time 2:

End Replication job (SnapMirror).

Time 3:

Modify data on primary.

The following file is removed:
/mnt1/B/file2.txt

Time 4:

Start Replication job (SnapVault created from SnapMirror source).

SnapVault incorrectly makes the copy from the SnapMirror source, and not the SnapMirror itself. Since the source no longer contains /mnt1/B/file2.txt, the SnapVault copy does not contain /mnt1/B/file2.txt.

In this example, the backup job (from the Backup From Snapshot operation) fails since a file was removed that was explicitly stated in the backup selections list. The backup job produces a status 71 – file doesn’t exist.
 

 




Article URL http://www.symantec.com/docs/TECH198753


Terms of use for this information are found in Legal Notices