Comprehensive data protection for all workloads
Post Reply
unsichtbarre
Service Provider
Posts: 226
Liked: 39 times
Joined: Mar 08, 2010 4:05 pm
Full Name: John Borhek
Contact:

Robocopy sanity check

Post by unsichtbarre »

Hi all,

I am migrating Backup Copy jobs from old disks to new disks to retire the older spindles. Mostly the repositories are configured ReFS 64K.

I'm using robocopy this way where e:\ is the old spindle and h:\ is the new spindle:

robocopy e:\backups\ h:\backups\ /J /MIR /MON:1 /MT:32 /r:1 /w:1 /copyall /b /timfix

I would expect robocopy to copy all of the data, remove files that were no longer present in the source (/MIR) (such as when Veeam applies retention policy) and get new files (MON:1) as they are created. Unfortunately, not only do I need to stop the Backup Copy job, but I need to break (Ctrl+C) and re-run robocopy in each instance. Where there are file-level backups, there can be a considerable delta, even if the first robocopy was done for many hours.

Is robocopy the best way to move to new disks? Am I missing something in my use of robocopy?

THX,
-JB
John Borhek, Solutions Architect
https://vmsources.com
Mildur
Product Manager
Posts: 8735
Liked: 2294 times
Joined: May 13, 2017 4:51 pm
Full Name: Fabian K.
Location: Switzerland
Contact:

Re: Robocopy sanity check

Post by Mildur »

If you want to preserve cloned blocks from refs (Fastclone), you need a block cloning tool to copy the disk content to your new backup storage.

With „normal file copy tools“ like robocopy, your fullbackups will be dehydrated. You need the exact space on the target storage from all your fulls and increments. Dehydrated backup files need longer time to copy because there is zero block cloning involved.
Product Management Analyst @ Veeam Software
unsichtbarre
Service Provider
Posts: 226
Liked: 39 times
Joined: Mar 08, 2010 4:05 pm
Full Name: John Borhek
Contact:

Re: Robocopy sanity check

Post by unsichtbarre »

I don't think an imaging or block-copy would work, unless I was moving to an identical size LUN?

For Example, if e:\ was a 20TB LUN, then f:\ would also have to be an identical 20TB LUN (possibly on newer spindles). Then, you might leverage dd or cat from a Linux image or partitioning tool (gparted) to move the data while the entire original disk was offline.

My goal is to move from older disks to newer ones, but also to consolidate numerous smaller repositories into single extents of up to 40TB or so. I think I am stuck with robocopy:
'
robocopy e:\backups\ h:\backups\ /J /MIR /MON:1 /MT:32 /r:1 /w:1 /copyall /b /timfix
John Borhek, Solutions Architect
https://vmsources.com
Mildur
Product Manager
Posts: 8735
Liked: 2294 times
Joined: May 13, 2017 4:51 pm
Full Name: Fabian K.
Location: Switzerland
Contact:

Re: Robocopy sanity check

Post by Mildur »

Ok, understandable.
I have never used robocopy todo file copy/sync.
I always use Fastcopy if I have to sync a fileserver disk to a new one. You have different modes. Sync Mode is what you are looking for, I think:
Sync Mode
Copy if size or date is different, or it does not exist. In addition, delete all destination files/directories that do not exist in Source.
https://fastcopy.jp/
Product Management Analyst @ Veeam Software
unsichtbarre
Service Provider
Posts: 226
Liked: 39 times
Joined: Mar 08, 2010 4:05 pm
Full Name: John Borhek
Contact:

Re: Robocopy sanity check

Post by unsichtbarre »

I will check Fastcopy out. Thanks!
John Borhek, Solutions Architect
https://vmsources.com
Post Reply

Who is online

Users browsing this forum: acmeconsulting, Semrush [Bot] and 110 guests