VMWare
ESX 5.5, Datastore on a HP 3PAR with 4x8GB FC (all paths active)
Backup Proxy
Windows 2008 R2 Server, 2 XEON CPUs, 64GB RAM
Datastores direct SAN attached with 2x4GB FC (all paths active)
Repository, 82TB (30TB free)
on Backup Proxy, 4x4GB attached (2 paths active)
We have a Windows 2008 R2 fileserver with 2x13.5TB disks as VM**. It's set to incremental forever, a full backup every month and backup files >16TB
The last successful backup generated 16.976.198.116kb
The backup did work for some time but when we increased the disks we started to get problems in veeam.
ERROR
Additional Details: 15.04.2015 15:49:11 :: Error: Not enough storage is available to process this command. Failed to write data to the file [V:\Backups\Fileserver\Fileserver2015-04-14T164737.vbk]. Failed to download disk. An existing connection was forcibly closed by the remote host Failed to upload disk. Agent failed to process method {DataTransfer.SyncDisk}.
Support
We were presented with some quite bogus solutions, like hot fixes for windows 2003. But no substantial help (fort time with veeam support so far).
Workaround 1
Nothing did help until I discovered myself that at the end of the backup only ~1GB RAM was available. So we upgraded to 64GB and it worked.
Error again
Well, it worked for about two weeks, then the same error started again. At the end of the backup >20GB of RAM were available. The server was online all the time, no ping lost.
Support
I opened a new case referring to the old one. No solution so far, but I get asked all the same questions again.
Things that I tried on my own but did not work
https://support.microsoft.com/en-us/kb/967351
https://www.microsoft.com/en-us/downloa ... px?id=9258
Set the Cache Size with powershell: http://serverfault.com/questions/325277 ... 466#527466
Workaround 2
Since the support did not come up with any solution and I need a backup of the VM I splitted the backup in two parts: OS and data disk 1 as one backup, data disk 2 as a second chained backup. This does work, BUT
Limitation 1: Failed to index guest file system. VSSControl: Index failed on the second data disk
Limitation 2: The backup with 2 disks runs at >300MB/sec. The backups with single disks run at ~160MB/sec
Limitation 3: Restore. I did not test this so far.
Solution?
It looks like that veeam cannot be used to backup big vmdks on a 2008 R2 repo. The problem is how veeam writes it backups file: one big file. This has to lead to some problems at the end of the day. If you think otherwise please feel welcome to share your ideas. At the moment I am really unhappy with veeam and their support since day after day the only thing the support does is to request the same information over and over.
If Windows is the problem than veeam has to point that out clearly that you cannot use windows servers as a repo for big file sizes. But since you need them for direct SAN attach (does not work with linux servers, or am I wrong?) and then you do not want to transfer your data over slow ethernet, I do not see how to handle this in any other way. It makes no sense to attach datastores and repo with FC and then use ethernet to connect them.
It would be no problem at all if veeam B&R just splits it's backup files since they already seem to know that windows servers have problems with such big files.
** before you tell me this is too big: this is our small document server. Our real data is on a 900TB beeGFS (Fraunhofer) backed up by TSM on 400 LT6. Welcome to the world of scientific research
