Backing up checkpoints to tape?

Hi Mike,

I understand your plan - however, instead of manually creating checkpoints and managing the backup sources, but if you are taking NDMP backups it will be much easier to use the Integrated Checkpoint with the NDMP Backup. When integrated snap option is used, a checkpoint is automatically created when the NDMP backup is initiated, the backup is taken from the checkpoint and once the backup is complete the checkpoint is automatically deleted.

I am not too familiar with NetBackup - but there must be a settings on the NetBackup side to use the snap while taking the backup.

Or else, you can set the NDMP snapsure=1 Parameter on the data mover to use the integrated checkpoint with the file system backup. Please refer to the Parameter guidelines or NDMP backup document for more details. Here is the summary of this parameter -

NDMP snapsure  0 or 1 Default:0

Determines whether EMC SnapSure is used for backup if the

NDMP environment variable SNAPSURE is not set or not

supported by the DMA software. The NDMP environment

variable, if set, overrides this parameter.

0 = Do not use SnapSure for backup.

1 = Use SnapSure for backup.

My 2 cents

Sandip

are you using NDMP?

Thanks for the quick reply.  We are using NDMP and we are employing Integrated Checkpoints with our other filesystems on the Celerra.  These specific filesystems in question, however, have special requirements.  To meet these requirements we are creating the manual checkpoints.  Furthermore, to meet requirements we need to be able to backup these checkpoints to tape and keep them for awhile.

I can do full backups of the checkpoint every evening, but I *really* don't want to - some of these filesystems will be rather large and I'd rather not "waste" the space in NetBackup if I don't need to.  That's why I'm going down this road of trying to find out why the Incremental backups are not behaving quite as I expected them to...

Thanks!

-Mike-

Yes.  We are using NDMP.

Hi Mike,

my *guess* is that the NDMP server is getting confused since the checkpoint you are asking incrementals for is different from the one with the full backup.

Have you tried to just refreshing pfs_current instead of moving and recreating it ?

Rainer 

I've not thought to try refreshing the checkpoint.  I will do that and see if the results are any different.  Thanks for the suggestion!

Rainer - that seems to have been the issue!  I ran a "-refresh" on the pfs_current filesystem and the incremental picked up only the changed files.  Now the question is how do I easily make a copy of the checkpoint just prior to running the "-refresh"?  Because to meet my requirements I will still need 30 days of individual daily checkpoints available to my customer.

Maybe I can use "nas_copy" to create the checkpoint copy just prior to running the refresh operation?

there is no way to copy a checkpoint unless you want a full clone

I would suggest to just create another ckpt to keep

Ah - that is precisely what I should do - create two checkpoints during my "checkpoint creation" script.  One to "keep" and the other a refresh.  I'll go down that road and see if it will do what I need it to do!

Thanks!

So the first time it does full ..then you run –refresh , does it create full backup after that ?

Hmmm... I've run into another oddity.  So I tried creating two checkpoints at the same time - one which will automount so the customer can browse the checkpoint directory, and the other (which I've named 'pfs_current') *not* automounted.  This "pfs_current" filesystem would be the one I would run the backups to tape on. 

Problem is, when I create this pfs_current checkpoint using "fs_ckpt pfs -name pfs_current -Create -option automount=no" and try running a NetBackup Full backup on it, nothing gets backed up.  I've tried creating this pfs_current filesystem *without* the "automount=no" option and a Full backup captures the entire filesystem.

So what am I running into with the "automount=no" option?  Why does that prevent me from backing up that checkpoint?  Any insight would be greatly appreciated!

a checkpoint needs to be mounted in order to access it or for backup. An unmounted checkpoint isn’t really useful

If you are familiar with Celerra take a look at /.etc/BackupDates in the data mover rootfs or have EMC support send you the contents – maybe that sheds some light on the behavior

Rainer