Repository is full....now what?

[Lyle]

I have 8 equal-sized SAN LUNs that our B&R 7 server can use for repositories. I though of striping them together at the Windows-level to give be one big repository, and the most flexibility, but Support said not to.

So for our 3 VMware clusters, I created 6 backup jobs, one for Windows VM's and one for Linux VM's (thinking to help deduplication), each with it's own LUN for storage. I know I should have done the math, or designed better, but I have no problem redoing the whole thing.

But one Repository is out of space. I can add a second repository for this group of VM's, but I guess that means splitting the job into 2 jobs, and moving files, or starting with fresh fulls or something. I've searched and read a few posts, but haven't really seen clear instructions on doing this. If someone could point me in the right direction it would help.

Or.... What about my original idea of one (or 3) big repositories, to make it less likely that I'll be shoveling VM's and jobs around in the future? Seems like if I did run out of space then, I could add more storage to the stripe instead of reconfiguring jobs.

Thanks....Lyle

[cvacanti]

need some more data...what OS is hosting the repositories? with new windows OS you can get the disks quite large...depending on esxi version too. I've seen a client go 4x2tb for repositories.

[foggy]

You can either just continue backing up to another repository without copying previous backup files (requires new full and the job will stop handling retention of backups in the old repository, you would need to do that manually) or expand the current repository. Splitting the job is also an option.

On a side note, there are built-in space notifications that allow to track free space on the repository and also Capacity Planning for Backup Repositories report in Veeam ONE that allows to monitor current space consumption and estimate the date when you run out of free space on repository.

[Lyle]

Thank you both for the input. My B&R server is a Windows 2008 R2 Standard SP1 64bit, with the single proxy and 6 repositories all on one ESX VM.

I'm still tempted by putting all six 800GB LUNS in one humungous striped volume and making one repository on that, just for ease of space management in the future. But maybe there would be an I/O bottleneck somewhere, or the stripe would be vulnerable some how (though it would be composed of "error-free" SAN LUNS).

Just following my (inexperienced) gut here, but rather than split jobs up (leading to less deduplication, I'd think) or redirecting the job output to a new, larger repository (and manually expiring the old repo), I deleted a couple of the repos I wasn't using yet, and used half of one to expand the Windows volume that was full. Then Windows saw a spanned volume with the increase in size, but the Repository was still the same size. Doing a "rescan" on the Repository let it expand to the full size of the newly increased volume.

So I think this is my answer for now.

Thanks again.....Lyle

[foggy]

Yep, that is probably the easiest way to address the issue. Thanks for the update.