How do you move a SV relationship between datasets in Protection Manager?

[reide]

IHAC who needs to move a single OSSV relationship and secondary volume out of one dataset and into another dataset.   Dataset "A" has 10 OSSV clients all using the same backup policy and schedule.  Each OSSV relationship has its own dedicated secondary volume (1:1 relationship).   The customer wants to move just one of these relationships to Dataset "B" so they can give it a different schedule.  However, any attempt to remove the OSSV client or its secondary volume from Dataset "A" results in Protection Manager deleting the relationship.  The customer wants to keep the relationship in-tact and avoid a re-baseline.

How can we accomplish this in Protection Manager?  Because each OSSV client has its own secondary volume, this should be relatively easy but we can't seem to get it to work.  I'm almost positive I've seen disucssions about how to do this, but I cannot find them.

Thanks.

[adaikkap · Accepted solution]

Hi Earls,

Here is what you need to do in Protection Manager

Step 1:Prevents PM’s reaper cleaning up any relationship.

Set the following options as below before doing the following and reset it back to orphans once done.

dfm options set dpReaperCleanupMode=Never

Step 2:Relinquish the Primary member and the secondary member.

· Use the dfpm dataset relinquish or NMC UI Edit Dataset Wizard.

Step3:Discovering as External Relationships.

· You must see this relationship as external in the External Relationship tab. If you don’t see it, close and re-login to NMC again,

Step4:Importing to a new dataset.

· Create a new dataset with required policy and schedule. Or choose the dataset where you want to import this relationship too.

· Use the Import wizard and import them.

Step 5:

· dfm options set dpReaperCleanupMode=orphans.

Points to take care:

1. If an entire OSSV host was added as a primary member, and now moved to a new dataset.(the step 2, relinquishing the primary member needs to be done for each dir/mntpath of the OSSV host.)

2. After importing the dynamic referencing of the OSSV host is lost as we import each individual relationships.

3. So when a new dir/mnt path is added to the OSSSV host, admin has to manually add it to the dataset.

4. To restore from Old backup version the use must go back to the old dataset as they are not moved over.

Regards

adai

[reide]

Thanks adai! That worked.

For anyone else who reads this discussion, some notes:

• When you relinquish the relationship, it may not show-up in the "External Relationship Lag" box of the Dashboard view. However, if you go to Data -> External Relationships it will be listed there.
  
• When you import the relationship into a new dataset, it will show an error status of "Baseline error". Simply run an on-demand backup job and it will clear this error. Note: The backup job doesn't perform a re-baseline. It simply does a Snapvault update.
  
• Don't delete the old dataset if its empty. As adai stated, it has the backup history of the relationship before you moved it. So if you want to perform a restore from before the move, you need to restore from the old dataset.   Once all the backups have expired from the old dataset, you can destroy it.
  

[emanuel]

Hello all

Can I apply this same technique when dealing with standard snapvault relationships?

I have a current dataset which has a number of haphazard relationships that is starting to fail - mainly conformance is detecting a deleted relationship for a specific element ( a qtree ) and the dpReBaselineMode is set to Confirm.

I need to expand this dataset to see exactly what is broken; the other warning is if I select Conform Now, it could re-baseline but if it does, will it attempt to re-baseline all my elements in this dataset?

thank you.

[adaikkap]

Yes its applies for any kind of relationship, SV,QSM,VSM,OSSV.

Yes when conform now is done it would do rebaseline for all rels that require, rebaseline.

Regards

adai

[kaikapola]

Hi,

I have dataset containing about 10 relationsships from 3 OSSV clients. When I tried to move one relationship to another dataset I got following error:

Cannot add secfil1:/vol/ossv_1 to node "Backup". secfile1:/vol/ossv_1 is already in dataset "OldDataset". A non-root member, or its ancestors or descendants, must not be direct members of any non-root nodes of any datasets.

Object, its ancestor or descendent is already in another dataset. (Error 23418).

This comes probably because I relinquished only primary member of relationship and don't relinquished secondary member (If I understand this right) even instructions says "Step 2:Relinquish the Primary member and the secondary member.".

Problem is that another 2 OSSV clients are still using same secfile1:/vol/ossv_1 volume for backups from different datasets and thats why I can't disconnect volume from "OldDataset". So, how I can accomplish to move relationship to another dataset without re-baseline?

Running 4.0.2.7803 (4.0.2).

Hope fully Netapp will make this easier in near future

Thnks, KK.

[paul_wolf]

KK,

I have the exact same situation. I have a dataset with 28 members and due to slow links, want to break that down into smaller data sets. I tried the exact same thing and got exactly the same error as you.

What is sounds like, is that you have to remove every primary in that dataset, remove the secondary, create new secondaries and import the primaries to the new secondary. Effectively, you are tombstoning the original secondary data set until all the old SV relationships age out and then you can remove the old secondary.

But I'm also interested in finding out for sure. I have opened a support case and when/if I get an answer I will post it here.

PRW

[adaikkap]

Hi KK,

            As you said, in-order to import the relationship you will have to relinquish both the source and destination. Also you a secondary volume cannot be a member of more than one dataset. But a primary volume can be. Pls follow the exact procedure to move the relationships.

Regards

adai

[kaikapola]

Hi,

I have come back again with this problem. I even opened case for this, but that was not successful operation.

I have two data sets SERVER1 and SERVER1_NEW and I'm trying to move all three relations from SERVER1 to SERVER1_NEW

Dataset SERVER1

C:\Users\me>dfpm dataset list -R SERVER1

Id         Name                        Policy                      Provisioning Policy Relationship Id State        Status  Hours Source                       Destination

---------- --------------------------- --------------------------- ------------------- --------------- ------------ ------- ----- ---------------------------- ----------------------------

    164207 SERVER1                    Policy_XXX                                               164215 snapvaulted  idle    18.2  SERVER1:C://                filerA:/SERVER1_backup/SERVER1_SERVER1_C___

    164207 SERVER1                    Policy_XXX                                               164216 snapvaulted  idle    18.2  SERVER1:SystemState         filerA:/SERVER1_backup/SERVER1_SERVER1_SystemState

    164207 SERVER1                    Policy_XXX                                               164217 snapvaulted  idle    18.2  SERVER1:E://                filerA:/SERVER1_backup/SERVER1_SERVER1_E___

Then I relinquished dataset:

C:\Users\me>dfpm dataset relinquish filerA:/SERVER1_backup/SERVER1_SERVER1_C___

Relinquished relationship (164215) with destination SERVER1_SERVER1_C___ (164213).

C:\Users\me>dfpm dataset relinquish filerA:/SERVER1_backup/SERVER1_SERVER1_E___

Relinquished relationship (164217) with destination SERVER1_SERVER1_E___ (164214).

C:\Users\me>dfpm dataset relinquish filerA:/SERVER1_backup/SERVER1_SERVER1_SystemState

Relinquished relationship (164216) with destination SERVER1_SERVER1_SystemState (164212).

Then I tried to import relations form Relations View. I have selected same three relations from above and got following error message:

=== CLIENT ===

Version=3.0.2.4201

Build=3.0.2

OS Name=Windows Server 2008

OS Arch=x86

OS Version=6.0

=== ERROR ===

com.netapp.nmf.userinterface.error.ResourceErrorDetails

=== MESSAGE ===

Cannot add filerA:/SERVER1_backup to node 'Backup'. filerA:/SERVER1_backup is already in dataset 'SERVER1'. A non-root member, or its ancestors or descendents, must not be direct members of any non-root nodes of any datasets.

=== DETAILS ===

Object, its ancestor or descendent is already in another dataset. (Error 23418)

=== CORRECTIVE ACTION ===

In order to add the dataset member or to add a DR protection to this dataset, you must first remove the member or its relatives from another dataset. Edit the primary node of the dataset to adjust dataset membership.

I tried same from CLI and got this:

C:\Users\me>dfpm dataset import SERVER1_NEW filerA:/SERVER1_backup/SERVER1_SERVER1_C___

Error: Could not import destination to dataset SERVER1_NEW (178629). Reason: Cannot add filerA:/SERVER1_backup to node 'Backup'. filerA:/SERVER1_backup is already in dataset 'SERVER1'. A non-root member, or its ancestors or descendents, must notbe direct members of any non-root nodes of any datasets.

I have feeling that I miss something, but can't figure out what? Command dpfm dataset relinquish can only accept destination as parameter, so how to relinquish source?

C:\Users\me>dfpm dataset relinquish help

NAME

    relinquish -- mark a relationship as external

SYNOPSIS

    dfpm dataset relinquish { [ <destination-volume-name-or-id> ] |

        [ <destination-qtree-name-or-id> ] }

DESCRIPTION

    The relationship will be marked as external. Source and

    destination objects are left unchanged.

Thanks, KK

[ARTSCHRAMM]

Did you ever figure this out?  I'm having the same issue. 

We have a large volume that I previously had a SM Dataset created for.  I've taken each LUN on the volume and placed it in it's own Qtree and now am attempting to create new datasets for a subset of the qtree's.  When I go to add the Protection Policy I get the following error message.  Is what I'm doing even possible?  Multiple Datasets for qtrees that exist on the same volume?

Thanks,

Art

=== CLIENT ===

Version=3.2.0.6001

Build=3.2

OS Name=Windows Server 2008 R2

OS Arch=x86

OS Version=6.1

=== ERROR ===

com.netapp.nmf.userinterface.error.ResourceErrorDetails

=== MESSAGE ===

Cannot set data protection policy to BI DW Data Local Snapshots to DR. SAN-PROD-03:/bidwdatap/dwdata is already in dataset 'BIDWDATA'. For a DR Mirror dataset, if the member is a qtree, then its siblings (e.g. other qtrees in the same volume with a member qtree) cannot be in the primary of any other datasets either.

=== DETAILS ===

Object, its ancestor or descendent is already in another dataset. (Error 23418)

=== CORRECTIVE ACTION ===

In order to add the dataset member or to add a DR protection to this dataset, you must first remove the member or its relatives from another dataset. Edit the primary node of the dataset to adjust dataset membership.

=== STACK TRACE ===

Cannot set data protection policy to BI DW Data Local Snapshots to DR. SAN-PROD-03:/bidwdatap/dwdata is already in dataset 'BIDWDATA'. For a DR Mirror dataset, if the member is a qtree, then its siblings (e.g. other qtrees in the same volume with a member qtree) cannot be in the primary of any other datasets either. (name=EDSCONFLICTWITHANOTHERDATASET,errno=23418,desc=Object, its ancestor or descendent is already in another dataset.)

[adaikkap]

Hi ART,

     SnapManager Dataset behaves little differently as they are mostly controller by the respective SM. If this had been a non-SnapManager or in NetApp term non-Application dataset, what you expect is possible.

Regards

adai

[adaikkap]

Hi KK,

     Sorry I missed one step. Once the relationships are relinquished you must remove the members in the following order primary members followed by secondary members.

Then if you try you import you should be able to do it.

Regards

adai