New backup job doesn't make use of mapped backup

[danielchung]

I was preparing a offsite backup job. According to the suggested solution, I made a local full backup, copy the vbm and vbk files to offsite, and then create a new backup job which is backing up from production site to offsite repository, also mapped the copied backup.

The new backup job runs but have a message " [vm name] is no longer processed by this job. Make sure this change is intentional". Also the job read and backup data as much as a full backup, so I believe it doesn't make use of the mapped backup and recognize the backup target as a new server.

I found the "cause" myself that is because in the first full local backup job, I pointed the target vm directly from ESXi host; where in the second full offsite backup job, I pointed that from vCenter. Knew that Veeam recognize VM by its unique ID assigned from ESXi host or vCenter, so that makes the job recognize the vm is no longer processed by the job.

You guys might say I should make another set of full backup and copy again. However, the backup size is about 100GB which I need 5 days to get it transferred to offsite over the slow link. I don't want to do it again. For some reasons I cannot login the ESXi host directly with my backup account, I must point to vCenter to do backup. So my question is, how to make the job recognize that the mapped backup is from the same vm that it's backing up? Also curious about how does the job know the vm in the backup is not the one it's backing up? By reading the vbm file? if so, wondering if modify the vbm file would workaround this...

[foggy]

Daniel, I'm afraid there's no way to change the VM ID, it is not stored anywhere in open format. So I think you either need to reseed using vCenter or get direct access to your ESXi host to point the new job at it.

[danielchung]

Alex, I've read some topics in the forum talking about the VM ID within the job needs to be updated after VM migrated from ESXi host/vCenter to other vCenter. It sounds similar to my case, say in day one, I only have a standalone ESXi host, backing up the VM locally by Veeam, and then after a while I migrate the ESXi host to be managed by a vCenter, then how to update the job so that it locates the VM from the vCenter? Wonder simply modify the job by re-adding the VM will lead exactly to my problem... or need to modify the Veeam DB?

I think this scenario is very common but somehow there is no simple way or even document about this. I've a support case 5181909, see if they can help me on this.

[ThomasMc]

As a side note you could read over the HOWTO: Backing up vCenter host issues thread in the powershell forum, this will give you an insight into what you would have to do, just be careful with it

[danielchung]

Thanks Thomas. Actually I've read through that thread and tried to figure out how to modify on which value will be helping me, but it's a bit complicated... looks having Microsoft SQL Server Management Studio Express will make it easier. But I still don't have confidence...

[ThomasMc]

the script will do what your looking for though, just change the VM name of the one your wanting to update, its not really just for vCenter you can do it to any vm you want(just change the VM name ), create a little test vm and try it out

[Added]
If you want to do if manually then this is sort of what you need to update

$VM = "vMachine.lab.local"
$NewStuff = Find-VBRViEntity -Name $VM | ?{$_.Reference -notlike "vm-*"}


In BObjects change
host_id =$NewStuff.ConnHost.Id
object_id =$NewStuff.Reference
path =$NewStuff.Path

[danielchung]

I'm not quite sure about some the codes.

Code: Select all

$vc = Find-VBRViEntity -Name $VCVMName | ?{$_.Reference -notlike "vm-*"}

I think this command is going to query my VM which is pointed to ESXi host (by criteria " -notlike "vm-*""), then store into variable $vc.

This is what I got from the $vc:
ConnHost : Veeam.Backup.Core.CHost
Type : Vc
Reference :
Id : 46bad7dd-165b-4cea-8161-cbcbb7513843
Name : MyVMName
Path : MyVMName

ConnHost : Veeam.Backup.Core.CHost
Type : Vm
Reference : 7
Id : 004964d3-a2b5-47d3-a63b-9105bba18990_7
Name : MyVMName
Path : MyESXiHost\MyVMName
FaultTolerance : False
GuestInfo : Veeam.Backup.Model.CGuestInfo
Uuid : 421bd7af-1e76-c54f-69cb-653876adf270
PowerState : PoweredOn

ConnHost : Veeam.Backup.Core.CHost
Type : Vc
Reference :
Id : 46bad7dd-165b-4cea-8161-cbcbb7513843
Name : MyVMName
Path : MyVMName


In the part of BOject change, it looks like to update the host_id, object_id and path with the value grabbed from $vc. However, when I try to read the value before putting into the SQL update, I got nothing neither from "$vc.ConnHost.Id", "$vc.Reference" or "$vc.Path". Also the query command update record where the VM is on ESXi host (by criteria "object_id NOT LIKE 'vm%'""). if i'm not wrong it won't actually update the record as value in $vc and that in the DB are the same.

Sorry that I'm not quite sure the script is helping me solving the problem. I guess if there is some ways to modify the vbm file, telling the job that the vm in the backup is pointed to vCenter might solve the problem. Read this thread but no solution yet http://forums.veeam.com/viewtopic.php?f=2&t=10540

[foggy]

I guess if there is some ways to modify the vbm file, telling the job that the vm in the backup is pointed to vCenter might solve the problem.

I can confirm that this cannot be done via editing the .vbm file.

[ThomasMc]

Ah, my mistake, you'll want to change the filter to like, I'll fire up the lab and get back to you with the exact commands, am sure I had to wrap them in $() to get the strings

[ThomasMc]

Code: Select all

$vcVM = Find-VBRViEntity -Name dc01.vpowercli.local | ?{$_.Reference -like "vm-*"}

"host_id = $($vcVM.ConnHost.Id)"
"object_id = $($vcVM.Reference)"
"path = $($vcVM.Path)"

[danielchung]

I failed to update host_id for my lab test machine, a sql error returned "Cannot insert duplicate key row in object 'dbo.bobjects' with unique index 'IX_BObjects'.", while the object_id and path are updated success. I queried the DB found that there is already a record of the test machine whose path pointed to vCenter and object_id is vm-xxx, so together with the modified record, 2 records in DB with different host_id.

I thought they will be removed automatically if there is no job or backup associated with the VM, so I removed the job, host and backup. However, the records are still there! did I make some steps wrong?

In addition, in the script, from the powershell command "find-vbrvientity ....", I got two identical results of the same VM. I think it's not normal but don't know what went wrong...

[ThomasMc]

Unfortunately no, the objects will be left behind in the BObjects table

[danielchung]

Thomas, do you mean that I have no way to fix it except rebuild the DB?

[ThomasMc]

There would be no need to rebuild the DB, it just means for some reason this was added to 2 jobs, one through the host and one through VC which is not what I was expecting as we wanted to change the original object in the DB instead of adding it to another job