floydburger

updated

13 years ago

0

2221

June 3rd, 2013 19:00

EMC Celerra DR testing

We're trying to come up with a valid way to test our replicated data from EMC Celerra without interupting production.

Setup

Version: 6.0.41-4, flare 30 on Clariion CX4-480

2.5TB CIFS file system associated with vdm1

Replication of both vdm1 and CIFS file system to remote site

The idea I had and have been trying is to create a checkpoint on the remote sites CIFS file system, create a new CIFS server and mount this checkpoint to this. The problem is vdm1 on the remote site is Unloaded, which pretty much rules out trying to do this.

We read that you could stop vdm1 replication from the production side, then load vdm1 on the remote site. By doing this, the productions CIFS server was then visible at the remote site and DNS was updated, which if we didn't change the record back it would have interupted connections to the production CIFS server.

This is the site we based the above test from

http://emcsan.wordpress.com/2011/11/29/testing-disaster-recovery-for-vdms-and-cifs-servers/

Thanks

  • 1075

    0

    Posted June 5th, 2013 00:00

    Fixed - the issue was I don't have permissions to the root directory of the checkpoint, i.e. bcptest, but I do have access to the subfolders. So I shouldn't try to share out /root_vdm_7/bcptest, I should try to share out /root_vdm_7/bcptest/data which the account has access to.

    [nasadmin@celerra ~]$ server_cifssupport server_2 -accessright -name admin -domain domain.local -path /root_vdm_7/bcptest/data
    server_2 : done

    ACCOUNT GENERAL INFORMATION

    Name                     : admin
    Domain                   : domain.local
    Path                     : /root_vdm_7/bcptest/data
    Allowed mask             : 0x301ff
    Action                   : List Folder / Read data
    Action                   : Create Files / Write data
    Action                   : Create Folders / Append Data
    Action                   : Read Extended Attributes
    Action                   : Write Extended Attributes
    Action                   : Traverse Folder / Execute File
    Action                   : Delete Subfolders and Files
    Action                   : Read Attributes
    Action                   : Write Attributes
    Action                   : Delete
    Action                   : Read Permissions

    1. Create test CIFS servers on DR side

    2. Create R/O checkpoint from the replicated file system on DR side

    3. Map a drive with your domain admin account to \\testcifsserver\c$ (this initiates a session so that your account is connected and the Celerra knows about you - without doing this, when running

    server_cifssupport server_2 -accessright -name admin -domain domain.local -path /root_vdm_7/bcptest

    it would fail to coonect to my account.

    3. Using MMC snapin computer management to test CIFS server

    4. Share out to c:\root_vdm_78\bcptest\data

    5. Map drive

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    688

    0

    Posted June 3rd, 2013 21:00

    create new VDM

  • 688

    0

    Posted June 3rd, 2013 21:00

    No it doesn't have to be the same CIFS server, ideally we'd want a completely new one. I've already tried this. On the DR side create a new CIFS server, join it to AD. Create a R/O or a R/W checkpoint (snapshot) but as soon as you go to mount it, it can't because vdm1 isn't loaded, or it can't see it becasue the replicated CIFS file system isn't associated with the new CIFS server.

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    688

    0

    Posted June 3rd, 2013 21:00

    does it have to be the same CIFS server, create a dummy one, create read/write snapshot and mount it to this CIFS server and verify data is there.

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    688

    0

    Posted June 3rd, 2013 21:00

    or create CIFS server on physical datamover, not virtual one

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    688

    0

    Posted June 3rd, 2013 22:00

    you are just creating a share:

    server_export server_2 -Protocol cifs -name myshare /root_vdm_1/ckpt_readonly

    you can see file system mount point using server_mount server_2

  • 688

    0

    Posted June 3rd, 2013 22:00

    Even when I create a new vdm (bcpvdm) and associate a new CIFS server to this bcpvdm, you still can't create a share off the checkpoint because it thinks it has no associated file systems. The key is with the checkpoint because when you view it, it's data mover is vdm1(R/O) and the mount is data mover vdm1. See below screen shots.

    cifs_cp.png

  • 688

    0

    Posted June 3rd, 2013 22:00

    ok - not sure how to do that. What does exporting do? Is this like mounting or associating the share?

  • 688

    0

    Posted June 3rd, 2013 22:00


    Ok, i've created a new CIFS server on physical datamover server_2

    Created a checkpoint of the replicated CIFS file system, using datamover server_2

    from this point i'm not sure what to do because if you go to Replicas, Mounts the checkpoint /bcptest is associated with datamover vdm1. This is because when you create the checkpoint even though you specify server_2 at the top, it knows it's associated with vdm1 - see pic below (Data Movers:)

    checkpoint.png

    The other issue is, now i've got the checkpoint and it's mounted you can't create a share from this to the new CIFS server because it doesn't see that the datamover has any associated CIFS servers - see below

    cifs.png

  • dynamox

    11 Legend

    •

    20419 Posts

    •

    87439 Points

    688

    0

    Posted June 3rd, 2013 22:00

    physical datamover can see that checkpoint, just need to export it ..use the command line.