Avamar with VMware backup /restore

Re: Vblock 1K Vm backup to AVE+DD

Hello Vijay,

Thanks for your answers.

You are right, the  customer has 2 isolated VLANs, so his idea is to have a seperate NIC for each. But i like the idea to have a AVE per VLAN, lets see if the customer likes it too ;-)

I don´t understand what you mean with "backing up VMs with NO network"?

What I meant was :  Avamar does not require the VM to have ANY network connection.

  The backup is completely independent of the network that the VM connects to.

Hi Tony -

I noticed you logged-in to the FTP site as "anonymous" , and I can see the directory you created, but I have the same issue - I cannot see the file name.

If you can re-upload and give me the file name, I can access that.

The more common way to upload is to login with the credentials  avamar_ftp/anonymous - note that the PASSWORD is anonymous. This puts the upload at a different location that I can browse.

Followed your instructions Vijay, still not sure if it has uploaded. File name is sadoc3_log.zip Hope you can find this.

Hello Tony:

I finally discovered where your uploaded files were hiding - after I found this piece of information on an internal site:

Due to a data security concerns, all files stored on ftp.emc.com/incoming  will be immediately moved to the same folder structure on and be accessible on the EMC network ... Anonymous access to ftp.emc.com/incoming is write only for files.

Anlyzing your file shows that your VM "sadoc3" has 2 virtual disks. Their Backup times and change rates are:

• sadoc3.vmdk:     20G  Change rate: 0.03% Backup time: 26 min
• sadoc3_1.vmdk 100G Change rate: 15.58% Backup time 3 hr 13 min.

The entire backup had 73599 changed  blocks.

On the 5.0 Avamar release, we have found that when there are over ~ 50 K changed blocks, there are diminishing returns to changed-block usage, particularly when there are a number of proxies loading the server with changed-block backups. This issue is particularly acute for the older Gen-3 Avamar hardware with 3TB nodes.

If you expect a continued 15% change rate on the second disk, I would recommend NOT using changed-block bakups for this VM.

Alternatively, backup this VM at a non-peak backup time, when there is not as much contention for processing changed-blocks.

Hello

We have problem with restore time of vm's. We have Avamar 6.0.66 and VMware 5.0 Essential,

so we do not have SCSI HOTADD.

SCSI HOTADD is only in enterprise vmware license.

SSLNBD transport is used. It is very slow due to SSL lib problem.

Is it possible (and how) to use not encrypted transport in avamar  - I think NBD is faster than SSL NBD.

You can use the flag:

    --transport=nbd:nbdssl

to specify that you want it to try "nbd" first, then "nbdssl".

This can be specified in either one of  2 ways:

• Edit the dataset for the vm, and add this flag: In the OPTIONS tab, click "MORE", then add the "attribute" "transport", and the "value" "nbd:nbdssl", and click "+", and save. (Preferred)
• For each proxy (assuming this is a 6.0 proxy. Path changes for 6.1), create or append to the file  /usr/local/avamar/var/avvcbimage.cmd the command "--transport=nbd:nbdssl" .

Yes - NBD should speed up the backup somewhat, although not as spectacular as hotadd.

What's up with Avamar not supporting file level restores from VM GPT or dynamic disks?  Neither are new and GPT in particular is becoming more of a standard as large VMs become more common.  Also, why does the Avamar backup job complete with no warning that file-level restore will not be possible?  Honestly, I'd expect more from a product from a company with a significant stake in VMware.

Also, why is mapping data stores to proxies such a pain?  Consider an environment where you have multiple large ESX clusters, each with their own set of datastores.  That is an awful lot of configuration and maintenance to keep datastore mapping up to date so backups use hotadd transport mode as often as possible.

Are there any plans to offer auto-discovery capabilities for VMs?  Perhaps something similar to the different type of affinities other products offer?

The File-level-restore (FLR) partition restrictions all come from vmWare's implementation limitations.

Avamar code basically uses the equivalent of vmware-mount, to mount the backed-up vmdk.

Since the restore restrictions are imposed by vmWare implementation (which may change), the Avamar software does not make a determination of FLR restore at backup time.

Re: Mapping data stores - I agree it could probably be rule-driven to make things easier. I will advise product management of your suffering.

Auto discovery of vm's is in an imminent release.

Hello guys,

I´m trying to perform a File Level Restore from a Image Backup. The problem is that fails after a few minutes with the following message:

avvmwfile Error <0000>: File level restore failed.

I´ve noticed that at the time that the backup fails, the proxy has already dismounted the VMDK.

I suppose there is a time out. If that is the case ... Where is it configured? Can I change this value?.

Thanks and regards.

Hi!

What permissions do you have set on the vcenter account?

Hi all.

I have problem with file level restore proxy. When I try to restore files from windows wm, the list of files is empty - without any allert or error.

Axionfs and samba works fine - I can access files from proxy.

Where can I find logs or something that can help me to check what is the problem?

In vcenter go to administration and roles

You should have created a custon role and it should have the following permissions.

DatStore

     Allocate Space

     Browse Datastore

     Low level file operations

     Move Datastore

     Remove Datastore

     Remove file

     Rename Datastore

Folder

     Create folder

Global

     Cancel Task

     Log event

     Settings

Network

     Assign network

Resource

     Assgin virtual machine resource pool

Sessions

     Validate sesion

Task

     Create

     Update

Virtual Machine

     Configuration

     Interaction

     Inventory

     Provision

     State

Vnetwork distributed switch

     create

     delete

     host operation

     modify

     move

     network IO control

     policy operation

     port config

     port setting

     vspan operation

Are there specific backups affected? Have you run a manual backup?