to cut a long story short: No, as of either ONTAP 7.3.4 or ONTAP 8.0, it is NOT possible.
Please get in touch with your local NetApp representative to issue a request why not or if possibly when such a feature will be available. NetApp wont give you a detailed, official answer in this community.
I'm having the same problems you face on my Netapp Filers. Have you any progress since end 2009 ?
I was wondering if you ever tried to remove the file from the filer using its command "rm" (rather than from NFS or CIFS mounts). I recently saw this command in "priv set advanced" mode and wonder if it could REALLY remove the file and not go into the snap.
By definition a snapshot is a read-only, point-in-time copy of a live file system, so deleting anything from a snapshot would a) require hacking into ONTAP/WAFL, b) break one of the fundamentals of data protection provided by snapshots.
There are ways to achieve similar goal without heavy cannons though:
- FlexClone the snapshot in question
- delete the file from the clone
- split the clone
- snapshot resulting volume (it now does not include the deleted file)
- delete the snapshot from the source volume
(yes, there will be ~100% capacity overhead after the operation)
I agree with you dmitri. last time when i tried to take the snapshot to tape it tokk me 15 hrs for dumping the snapshot to tape which is very tough for us to do this for all snapshots & re-claim the space back.
>Is it possible to restructure your data into separate volumes so that the data you don't want to be included in a snapshot is stored on a volume without a snapshot schedule or data >protection policy?
I mentioned in my original post that we play these games now. I find the status quo unacceptable. There's still the chance that someone accidentally creates a large file where they shouldn't have and the admin has no way to deal with that problem.
>If you add the ability to delete files out of a snapshot how do you guarantee the consistency of the snapshot contents?
How do you guarantee the consistency of any filesystem? There's a whole branch of OS theory to deal with these issues.
Don't you think that's a little bit ridiculous? We should lean on Netapp to give us a little bit more power over the filesystem. Make it a privileged command if you want just to be really, really safe. I should not have to kill 800,000 files I'd like to retain snapshots of just because there is 1 (large) file in the snapshot I don't need which I'd like to remove. I am not fully aware if there are technical reasons why this isn't implemented, but it seems like it would be a simple matter of adding the blocks to the free list and updating the master blocks which contain snapshot info. I would *really* like to see this.