Gotchas or issues wiping Aux Copied Storage to reconfigure mount points

I think this simplest way to accomplish this is to delete the secondary copy associated with the source of the tape copy and that 256 TB mount point within commvault and allow Commvault to clean out the data. This will take care of the deduplication database which would otherwise be a bit tricky to deal with -- since it wont know the data is gone and would pose a risk of deduplicating against blocks that are no longer there. It might take some time to process ¼ of a PB of data, but you can immediately create a new copy with your new mount points and start transferring data while the deletion is happening.

Summary, if I understood your scenario.

1. Delete the copy pointing to the 256TB library
2. Run data aging manually
3. Create your new library with the new mount points
4. Create a new copy associated with your new library -- associate with the existing global DDB you mentioned
5. Set the source of the new copy to your primary (source site)
6. Adjust the tape copy to copy from the source of your new copy

Hi ​@tigger2,

The points Damian mentions are correct but I think it is a bit sparse and I would do some stuff differently. If you are going to delete all data and change important stuff like this, it is better to start over for all components.

I would advice taking these steps. I don't know if you have multiple storage policies so repeat all steps concerning changing, deleting and creating copies for all necessary storage policies.

1. Make sure aux copy to tape is all done.
2. Change the source copy for tape to the primary copy. This way you tape copies can still run while you are deleting, recreating and refilling the secondary disk copy.
3. Delete the secondary disk copy. Make sure to note down settings such as retention before deleting.
4. After all copies are deleted, delete storage pool(s) associated to the disk library. If this is an old config, you may need to delete the global deduplication storage policy to get rid of the DDB association.
5. Delete disk library. You may want to stop the services on the MA so it will only be removed from the CV config as this is faster and you will delete the entire partition anyways.
6. Delete mount point partition on the Windows Media Agent.
7. Delete DDB folder to empty out DDB disk.
8. If needed, change mounted LUN sizing.
9. Create new partitions and mount them on te server. I would advice creating a DiskLibrary folder where you create mount folders for the partitions. This makes it easier to exclude the mount paths from AV software scans.
10. Create a new Storage pool. This will create the DDB policy and a disk library. Add mount paths to the disk library after creation. If needed make changes to the config.
11. Create a new secondary disk copy with same settings as before. Make sure that you choose All backups to be copied. Because a lot of data will be copied, Commvault will ask that you change the jobs to copied starting date to a more recent date but you can decline this.
12. Run aux copies. If needed create a temporary aux copy schedule.
13. Once aux copies have been performed, you verify that the storage contains the amount of jobs and application size you expect. If not make sure these jobs are selected.
14. Once all data has been copied, change the source copy for your tape copies back to the new secondary disk copy.

After this you should be able to work as before with the renewed disk library.

Kind regards,

Paul

Everyone: Thanks for the info! 

I’ve never messed with storage/DDB/or copy policies (just maintaining what was left from previous admin), and “when this is to be done” I’ll prob open a ticket to officially “CYA” for this type of work, but I wanted an idea of what it would take to plan for options (also in case it was not do-able without serious danger/massive reconfig/time constraints) as support tends to be… sometimes … no so creative with solutions and takes a lot of time to back and forth with them.

If I understand things correctly: The idea seems “do-able”!

​@Damian Andre: It appears your solution is to let commvault clean up the storage itself and part of the issue I have doing this (which I did not mention initially) is the “Create your new library with the new mount points” because… there is no other library (or that much unallocated space) to add this much additional space to 😥.  We do have *some* extra unallocated space (and could just make another small mount point with it) but will likely have to buy another library/expansion, which we hope to do that when we do a hardware refresh in a year+.  Also (I did not mention): there are several TB’s of unallocated space on the backend storage accidentally added to the 256 TB LUN (the LUN is larger than we could allocate in windows...oops), so deleting the 256 TB mount point entirely will free up extra space we need to allocate to the new (smaller) mount points we hope to make. We do have the option to buy another set of storage arrays for the needed space, but that hardware refresh date is a little too close to want spend $ on new storage arrays if we can avoid it until then.

​@Paul G : I think we will prob. end up going this route as we wanted to delete the mount points/storage entirely and start over with repartitioned? storage/mount points immediately...for reasons (see response to Damian above for things I did not mention in my initial question).  We might have a bit of an issue if we have to run from tapes on our primary copy (due to slowness likely happening) BUT I think if we plan to do this quickly, and in 1-2 days, we can get it all set up (and allll the data copied back...another week+ prob) before the next round of tapes need to run.

Hi ​@tigger2,

Your secondary disk copy will also need to read all data from the other disk copy to recopy it to the new disk library. If this storage cannot handle the load of both the disk and tape copies reading at the same time, I would advice disabling the aux copy schedule for the tape copy until your secondary disk library has been rebuild and filled. It is better to have your backups running to 1 copy than none and jobs will be retained until copied to tape.

Also, if 2 copies reading from storage is enough to cause issues with the disk storage, I would expect longer copy times. I usually expect 100-150 TB a week for aux copy for disk to disk copies in a situation where a copy needs to be rebuild but this is highly dependent on the type of data and setup. So don't expect this to be finished within a few days as it is not just 250 TB of data but probably over a petabyte of application data which all needs to be processed.

I hope this made things clearer for you. If you are not confident about these actions, make sure you get support or professional services to help you. It is better to take a bit longer with help than accidentally deleting all data or creating a situation where no backups can run anymore.

Kind regards,

Paul