Help me understand requirements for Nimble snapshot integration

[rrkerr]

Hi all,
I've read blog posts, the user guide and integration guides (Veeam) but have something wrong as I keep seeing messages to the effect "no
suitable proxy found for storage integration, reverting to VMware snapshots".

Backup job logs don't reveal anything I can make sense of other than "suitable proxy not found"

In summary the HPE dHCI primary infrastructure comprises:

Proliant DL325 with 10GbE
vSphere 7.02 Ent Plus cluster
Nimble AF iSCSI 10GbE
Veeam v11 Ent Plus

I'm simply trying to leverage Nimble snapshots within Veeam Backup Jobs for VMs located in the primary cluster.

Nimble is added to Veeam Storage Infrastructure successfully and can be explored through Veeam GUI
Proxies are multi-homed Windows 2016 VMs, located in the source cluster with network connectivity to one of the Nimble iSCSI networks. iSCSI target discovery seems good.
IQN of proxy VM added to Nimble initiator group
Backup repository is configured
Backup job is set to use only the proxy configured with Nimble connectivity
The backup proxy transport mode is Automatic

It's not clear to me if a virtual Windows proxy requires NCM software to be installed or not. Even although I'm using a virtual proxy, should it setup in the same way as if a physical server with direct SAN attach; I thought not as my proxy would use hot-add mode?

I was expecting the Veeam backup job to initiate a storage snapshot, the virtual proxy to mount the snapshot volume and then use hot-add method to access individual VM disks for backup. Instead the storage snapshot isn't initiated because a suitable proxy can't be found, and the backup completes using VMware snapshot instead. What am I missing/got wrong? Help and tips appreciated! Please don't point me to a guide as I'm obviously already misunderstanding something fundamental here. I know it should work so some specific configuration pointers needed I think.. thanks.

[rennerstefan]

Hi,

thanks for reaching out and sorry to hear it is not yet working after reading through everything.
First I'd like to ask you to open a support ticket as it is always good to have our team checking the logs in such a case.

Now to what you wrote:
- Proxies seems to be ok if the are available in the management network the Veeam Server belongs to and also have a interface to to iSCSI
- IQN handling would be done by us but that should not matter
- Repo doesn't matter at this point
- transport mode is ok as we will use storage snapshots if possible

You are also right that the virtual proxy is more or less configured as it would be a physical machine.

One thing that you are wrong in is the process so let me explain a little:
1. We will check if the VM you try to backup is on a Nimble LUN that we can access (seems here is already the issue)
2. We will trigger a VMware Snapshot followed by a Nimble Snapshot. The VMware Snapshot will be removed once the Nimble Snapshot is created.
3. We will mount the Snapshot to the proxy via iSCSI
4. The Proxy will read from the Snapshot and write to the Repo (No Hot add is used in this case, it is direct from Snapshot to Repo)
5. Once Backup is done the Nimble Snapshot will be deleted

As this is not working for you and based on the message you provided the issue seems to be that either the proxy can't access the Nimble on the Data Interface or the VM you try to backup is not on a LUN we can access.
That said, please check once again that everything is working with your proxy to nimble connection.
What you can try is, create a manuel Snapshot of a LUN on the Nimble, perform a rescan in Veeam of that LUN and check if you can see the VMs that are supposed to be in that LUN. With that you would know if the proxy is able to scan the nimble which is essential for the backup to run. I recommend to do so by creating a Temporary LUN and VM so no impact to production can happen while you do it.

For everything else a support case and logs would be required to check what may be the issue.

Let us know.

Thanks

[Andreas Neufert]

Check the "History"-"System" area if the rescans work correctly.

Not related to this issue, but please make sure that the access policy for the VMware Servers are set to "Volume only" and not Volume+Snapshots" to avoid issues over time.
The access policy for the Veeam Proxy server has to be volume+snapshots but I guess we create the Host entry and set it accrodingly.

[rrkerr]

Thanks Stefan

You are also right that the virtual proxy is more or less configured as it would be a physical machine.

I'm missing something here - I get that a physical proxy would use direct SAN access and actually mount the VMFS volumes where the VM resides, correct? This means for Nimble storage that physical proxy would need the Nimble Connection Manager software installed.

You state that the in the process, Veeam proxy mounts the snapshot volume, not the actual volume, and reads the data using direct SAN method (not hot-add).
This implies that the virtual proxy will indeed need the NCM software, and be explicitly granted access to the snapshot volumes? How is that possible if the snapshot doesn't actually exist until time of backup? How is the virtual proxy granted access to only the snapshot volume at time of backup?
Does the virtual proxy need access to all the VMFS volumes as well? That sounds risky as a virtual server has direct SAN access to volumes it's own cluster host owns.

I can confirm the VM resides on Nimble LUNs only.

Andreas - thanks also...
I've attempted a snapshot back from within Veeam and it fails with a similar message. Haven't been able to the rescan function yet though after taking a snap using Nimble GUI. I can see the volumes and historic scheduled snapshots but didn't think to check the VMs present in the volume.
Where is the access policy you refer to - can you elaborate as I think this might be part of the missing piece of configuration?

Thanks

[Andreas Neufert]

The official configuration guidelines from Nimble highlight the setting "volume only" for VMware processing in the official documentation.
Nimble KB-000367
It can be accessed over this HPE account webpage:
https://infosight.hpe.com/InfoSight/med ... family.pdf

[rennerstefan]

Hi again,

correct. The proxy will mount the snapshot to read data out of it using backup from storage snapshots.

That means, we (Veeam VBR) will call the Nimble API (as it is added to VBR) to create the Snapshot and then perform the mounting incl. the handling of the access requirements on nimble. That said, you don't need to care about it as we will do it.
NO, the proxy does not need access to the live volumes, as you said this is to risky and the reason to use Backup from Storage Snapshots instead of Direct SAN.

So as mentioned before it is all about the proxy being able to access to nimbles Data Interfaces (not the actual live luns).

Thanks

[rrkerr]

Okay - with the exception of NCM software question, I think I meet all the requirements. It's a bit ambiguous in documentation but the Nimble doc does say that the NCM software should be installed on the proxy, however the context is using a Nimble volume as a repository, combined with proxy role. My repo is on another infrastructure connected.

In my scenario, if the virtual proxy mounts the snapshot volume and Veeam handles all the access requirements, does that mean NCM is NOT required to be installed?
I can't see how NCM is needed as the virtual proxy isn't mounting a "real" LUN, only snapshot. If I'm understanding correctly, then I think something isn't working as expected and my next step is to escalate with Support.


Thanks again!

[rennerstefan]

As far as I know NCM is not required on the proxy.
Yes as already mentioned open a case for sure and also post the case ID here.