NetBackup 7.5 + VMware 5.0 making full backups instead of Differential-inc

[therockyb]

Hi,

We have been using Netbackup 7.5 since First Availability and switched all our policies for Vmware from FlashBackup-Windows to Vmware.

Master and Media servers are running Windows 2008R2 all at Netbackup version 7.5 and all our vcenter servers are running vSpehere 5.

We are having problem with differential-inc backups on our Vmware policies; the differential-inc backups are the same size of the full backup.

One thing I noticed is the file count, on full backups the file count show all the files but on differential-inc backups I see a fragment of the files of the full backup which makes me think that the file count is actually the good one. Both Full and Diff are taking around the same time to complete.

I went through the different articles:

- Validated my permissions making sure that ''disk change tracking'' is enabled on the role ''VMware Consolidated Backup User''

- Validated my ''change block tracking'' it is activated and the VM level and at the disk level

- My servers are all in the same policies for full and differential, I made sure to run a full backup first

I am at the point where I don`t know what to check anymore, any of you would have ideas?

Thank you

[Marianne]

Have you selected "Perform block level incremental backups" in the policy attributes?

[Yasuhisa_Ishika]

Please check following points. * (Policy) Block Level incremental backup or not * (Policy) mapped full VM backup or full VM backup * (VMware Backup Host) Incrementals based on timestamp or incrementals based on archive bit * (VMware datastore) what type of datastore, SAN/DAS/iSCSI or NFS?

[tarangdce]

I had a same issue where i took 1st backup without BLI selected and next incremental with BLI selected.

I copy the same policy into new with BLI selected.

Run a FULL backup and subsequent Incrementals are quite less.

[Mark_Solutions]

Remeber that if you make a substantial change to a policy (chaging from FlashBackup-Windows to VMWare or from BLIB to no BLIB) then you need to run a first Full backup with the new setting otherwise all Incremental backups will run as Full until the next Full has actually run

Hope this helps

[therockyb]

Hi,

I should have started with the basics; yes ''perform block level incremental backups'' is selected, in fact the policy is set to ''VMware'' which grey out this section and select it by default. The policy is also set for ''Enable file recovery from VM backup'' and ''Enable block-level incremental backup''.

The datastore is SAN and the transport is also SAN.

I don`t think that the ''incrementals based on timestamp or incrementals based on archive bit'' is available anymore, I believe it was replaced by ''change block tracking''.

Those policies were transfered from FlashBackup-Windows to VMware 4 weeks ago, I installed 7.5 the same day first availability was released, I performed at least 3 to 4 Full backup since then.

Any other ideas?

[VirtualED]

This is a problem with your particular VM.  For whatever the reason all the blocks marked on the Full and being mark again on the differential.  I would check for Orphaned Snapshots on the VM.  Cleanup the VM Snapshots, maybe vRanger the VM and see if that addresses the issue.

What OS is the VM?

[therockyb]

Hi VirtualED, I am having the problem with all my VMs, Windows 2003, 2008R2, CentOS and RedHat.

I would say 90% of our VMs were never snapshoted except from the snapshot that Netbackup generates.

[Yasuhisa_Ishika]

Have you already tried disabling and enabling CBT? If not, consider to try as below. 1) Shutdown VM 2) remove all snapshot of VM 3) remove CBT setting 4) remove *-ctk.vmdk files on datastore 5) enable CBT for both VM and vmdk 6) power on VM 7) take full backup 8) take diff backup

[therockyb]

Hi,

I am still working on this problem trying to find out why my differential backup are doing full backup.

I contacted the technical support but I ended up talking with a tech without any knowledge on backup specific to VMware therefore the case is not progressing.

I am starting to suspect two things:

1 - Problem is with policies that are using a client selection ''Select Automatically through query''.

2 - Problem is with policies that are not using the ''VM hostname'' has the ''Primary VM identifier''.

Is there anybody using client selection through query and/or primary VM identifier not being VM hostname and having successful differential backup?

[therockyb]

Hi,

I wanted to follow-up on this issue, I received a call today from Symantec and they confirmed that differentiel backups are not working with Netbackup 7.5 and vSphere 5 if you are using client selection ''Select Automatically through query''; the problem was reported by multiple customers and they were able to reproduce the problem internally.

They are currently working on a fix but there is no target date available.

I will follow-up once I receive the patch from Symantec.

[julmer]

Hi,

is there any update / fix on this issue?

Kind regards
Julian

[therockyb]

Hi Julmer,

There is still no fix for this problem.

They are following up every week with me to let me know that they are working on it but at the moment there is still no resolution.

[Omar_Villa]

If you have any DB's in your boxes and you have the special options enabled (Exchange, SQL, SharePoint or any DB) they will perform a full backup all the time even when the policy is set to differential or incremental because of the VSS snapshot process, not sure if this is the case or if maybe you have the bug they found plus some boxes under this configuration.

[therockyb]

Hi Omar,

This bug is occuring without any application protection enabled on the VMware policy.

[julmer]

Hi therockyb,

so I would appreciate if you post on this thread if there is a solution. I'll keep up watching it day for day.

Kind regards
Julian

[Bmitche]

Anyone know if this has been fixed? I would hate to place a service call and go through 4hrs of try this try that and send me 40gb of logs....only for the tech to say, "it's a known issue we will get back with you".

[therockyb]

Hi Bmitche,

Still no fix, just received an update today from Symantec, <<Engineering continues to work on this issue...I hope to have something for you in the next week.  If I have something promising I'll update you immediately; otherwise I'll check in again next Tuesday.>>

[DF-FAPS]

We have 2 of the 5220 appliances both configured as master servers. The VMWare ESX version on most of our hosts is 4.0 and 4.1. We just performed an appliance upgrade from 2.0.3 to 2.5. After the upgrade several VMs ran a differential backup. The strange part was that they took just as long as the previous fulls a few days ago prior to the upgrade. They were just as large as the last full backup. Even stranger is that the file count was what a normal differential should be.

So, a differential job with differential file count, running like a full in both time and data size. Additionally the dedup rate went from the 90% range to 50% or less. The dedup pool used up several extra terabytes.

[therockyb · Accepted solution]

Hi,

I want to report that this bug is corrected.

Symantec sent me the binaries, I tested them and it is now working correctly.

Please see the following tech note that was created related to this fix.

TECH189975

http://www.symantec.com/business/support/index?page=content&id=TECH189975