Direct Access Backup with FC SAN

[sadeghalmahdi]

Hi Guys
I have some problem and confused.
I decide Backup Via Veeam Program Through FC And My Topology Described Bellow:
one EMC VNX 5200 SAN Storage Connected To My 3 ESXi Host Via SAN Switch.
One Of My ESXi Host Is for Backup Solution And Veeam Backup Replication Software Is Virtual Machine And Installed on Server 2012.
In My Backup Host all of EMC SAN Storage LUN Mapped and another SAN Storage Attached Directly to Me Veeam VM via RDM.
but I Can not Directly Backup From SAN Storage When I Set Backup Proxy Advanced Option in SAN Direct Storage Access.
Can I Backup Directly From SAN Storage When My Veeam Backup Server And Proxy Is Virtual Machine.
Thank You

[PTide]

Hi,

Can I Backup Directly From SAN Storage When My Veeam Backup Server And Proxy Is Virtual Machine.

Yes, you can. Do they occupy the same VM or separate VMs? Your backup proxy has to have an access to LUNs where the VMs that are subjects to backup reside at. Please check the Requirements for the Direct SAN Access Mode part of this article and make sure that your backup infrastructure meets those requirements.

Thanks.

[jumave]

We want to use a virtual Veeam proxy server running on a dedicated backup vSphere Cluster.
The VMDK storages containing the VMs are situated on LUNs on a storage system connected to the host by a FC SAN fabric.
As far as I understand it should be possible to present the LUNs to the proxy server VM by means of RDM.
Is it a supported scenario to access the VMDK storages by means of direct SAN access with the border conditions given above?

[foggy]

We want to use a virtual Veeam proxy server running on a dedicated backup vSphere Cluster.

Is this a separate cluster, not the one containing the backed up VMs?

[jumave]

It is a separate cluster

[PTide]

Hi,

As far as I understand it should be possible to present the LUNs to the proxy server VM by means of RDM.

I'm not sure if it's possible to add LUN being used by host as RDM, that's a subject to test, have you tried that already? Do you have any VMs running from FC SAN on the backup cluster?

Thanks

[mma]

Hi

Since when is it possible to do direct SAN backups on vSphere from a virtual proxy?
I would say direct SAN on vSphere needs a physical proxy.

But let me know if I am wrong

[PTide]

It is possible to use VM proxy for direct SAN, although it is recommended to stick with physical one.

Thanks

[naugul]

Hi!

Let me start by apologizing, i'm sure the answers are out there, somewhere, but i'm tired of looking, everything i find seems to be outdated...
I'm new to Veeam and to SANs and i'm pulling my hairs out by now...
I'm trying to come up with the best backup solution for a client, currently their infrastructure looks like this:

1 x Dell SC4020
2 x Dell R630 (4 x SUSE VMs on each)
1 x Dell T330 (2 x Windows Server 2016 VM, one of which has the Veeam B&R)

The 3 servers are connected to the SC4020 through FC, there is one LUN for each server, and every ESXi host can access any LUN.

In my head, the best option would be to "direct access" the SAN from the Virtualized Veeam Server, but i can't seem to understand how to do that. In the meantime, i tried to implement proxys on each host and run the backups using them in Virtual Appliance mode, but when the VMs are to large and the tasks goes for over 3 or 4 hours something crashes, sometimes the proxy, sometimes the entire ESXi host.

So... Is it really possible, and if so, how can i give the Veeam access to the VMFS LUNs?

Thanks!
Diego.

[Vitaliy S.]

Hi Diego,

In my head, the best option would be to "direct access" the SAN from the Virtualized Veeam Server, but i can't seem to understand how to do that.

Direct SAN access is mostly recommended when using physical proxy servers. In this case you separate your backup infrastructure and production VMs from each other. Using SAN mode with virtual proxy servers usually doesn't bring any benefits over hot add mode.

Anyway, to use a direct SAN mode you need to present LUNs via FC to the VM. The process doesn't differ from presenting LUNs to the physical proxy servers.

In the meantime, i tried to implement proxys on each host and run the backups using them in Virtual Appliance mode, but when the VMs are too large and the tasks goes for over 3 or 4 hours something crashes, sometimes the proxy, sometimes the entire ESXi host.

What is the error message in this case? What are the bottleneck stats for your backup jobs?

Thanks!

[naugul]

Hi Vitaliy!

Direct SAN access is mostly recommended when using physical proxy servers. In this case you separate your backup infrastructure and production VMs from each other. Using SAN mode with virtual proxy servers usually doesn't bring any benefits over hot add mode.

Ok, lets just say that i want to try anyway... (The average speed of my backups, using Virtual Appliance Proxys, is about 20/40 MBs... And between the 8 VMs i've got 8TB of data... Every Active Full is a pain...)... Given that the Full VBR Server has direct access to the SAN via FC, i certainly think it should be faster...

Anyway, to use a direct SAN mode you need to present LUNs via FC to the VM. The process doesn't differ from presenting LUNs to the physical proxy servers.

I'm sorry, i get that, and again, sorry... But how do you do that? I mean, when i did it with the physical servers the FC switch showed all the pertinent WWNs, i created the masks, configured the SAN so that the LUN could be mapped on it, etc... it was straightforward... But not with the VM... I tried VPID, couldn't get anywhere, and anyway, is it really necessary? I need the VM to access all the LUNs the ESXi Host can access...
I realize this is more of a question to VMware Pros than Veeam Pros ( ) but i'm kind of hoping there is a button somewhere in vSphere that would just "share" the VMFS volumes with the VM. (Already tried RDM, it doesn't show the LUNs).
I also thought about "Shared PCI Device", but i don't think that's the way to go...

What is the error message in this case? What are the bottleneck stats for your backup jobs?

The error message is: "'Veeam.Backup.AgentProvider.AgentClosedException'"
The bottlenecks are always "Source".

I've finally manage to successfully finish the job in the last host, the one with the most data in the VMs. I had to create a job with only one of the VMs, run it, wait for it to end (about 5 hs), edit it, add another VM, run it again (so that the previous VM would finish almost instantly, wait another 5hs or so), edit it, add another, etc... As long as the backup isn't a full backup, the job runs fine now.

It looks like something along the way is running out of resources... Should i give the proxy more processing power? (It currently has 4 E5-2960 CPUs @2.6ghz. and 4GB of RAM - According to vSphere, it never uses more than 50%) The weird thing is that most of the times, when it fails and gives that error, i need to do a cold restart of the host because it becomes unresponsive... it won't even answer pings!

Thanks for your help!

[Vitaliy S.]

Ok, lets just say that i want to try anyway... (The average speed of my backups, using Virtual Appliance Proxys, is about 20/40 MBs... And between the 8 VMs i've got 8TB of data... Every Active Full is a pain...)... Given that the Full VBR Server has direct access to the SAN via FC, i certainly think it should be faster...

The bottlenecks are always "Source".

If you job reports source as a bottleneck, then you proxy server performs just fine. Can you please confirm that hotadd mode was used? You can do that by reviewing the corresponding job session log.

It looks like something along the way is running out of resources... Should i give the proxy more processing power? (It currently has 4 E5-2960 CPUs @2.6ghz. and 4GB of RAM - According to vSphere, it never uses more than 50%) The weird thing is that most of the times, when it fails and gives that error, i need to do a cold restart of the host because it becomes unresponsive... it won't even answer pings!

When the host crushes, do you see any errors in the tasks & events log of the ESXi host?

I'm sorry, i get that, and again, sorry... But how do you do that?

I guess NPIV should be the solution in your case.

[naugul]

Hi Vitaly!

I'm bringing this back to life... Sorry... I never got to answer you...

If you job reports source as a bottleneck, then you proxy server performs just fine.

Ok!

When the host crushes, do you see any errors in the tasks & events log of the ESXi host?

Yes, there are. It turns out the crash is a problem in the ESXi Lan Driver for some Dell Servers... In my case, Dell R630, particularly, the NTG3 driver.
I got around that, originally, by backing up one vm at a time in a Backup Proxy on each host and just transferring the compressed data through the network. After the first backup, i had no problem, being all the next backups just incremental...

I guess NPIV should be the solution in your case.

Now, regarding NPIV, i never could access that site, it says i'm not authorized to read that forum.

___________________________

Now i'm really complicated. I need to perform New Full Backups, and my VMs are way more heavy now... Because of the problem with the NIC driver, still unresolved, i definitively need to figure out how to make Virtual Appliance or Direct Storage Access work...

Can you give me a hand?

If i try Direct Storage i get this error:

Code: Select all

Error: Failed to open VDDK disk [[SAP_DEV-QAS] TestWindows/TestWindows.vmdk] ( is read-only mode - [true] ) Logon attempt with parameters [VC/ESX: [172.16.0.20];Port: 443;Login: [administrator@iturria.com.ar];VMX Spec: [moref=vm-72];Snapshot mor: [snapshot-74];Transports: [san];Read Only: [true]] failed because of the following errors: Failed to open disk for read. Failed to upload disk. Agent failed to process method {DataTransfer.SyncDisk}. 

Since i never could make the VBR VM have direct access to the VMFS datastore i'm assuming the problem comes from there...

Now, if i try Virtual Appliance i get this:

Code: Select all

Preparing backup proxy VMware Backup Proxy for disk Hard disk 1 [hotadd]
Error: Invoke failed. Command: 'HotAdd.AttachDisks', parameters: {(EStringArray) vddkPaths=[vddk://<vddkConnSpec><viConn name=~=172.16.0.20~= authdPort=~=443~= vicPort=~=443~= /><vmxPath vmRef=~=vm-72~= datacenterRef=~=datacenter-2~= datacenterInventoryPath=~=Iturria S.A.~= snapshotRef=~=snapshot-73~= datastoreName=~=SAP_DEV-QAS~= path=~=TestWindows/TestWindows.vmx~= /><vmdkPath datastoreName=~=SAP_DEV-QAS~= path=~=TestWindows/TestWindows.vmdk~= /><transports seq=~=hotadd~= nbdCompressionType=~=none~= /><re

Here i really don't get what the problem could be... I have complete access to all datastores from any ESX host... And, i've tried this -> https://www.veeam.com/kb1184 and it works perfectly...

Any ideas?

Thanks!
Diego.

[Vitaliy S.]

Hi Diego,

Now, regarding NPIV, i never could access that site, it says i'm not authorized to read that forum.

My bad, it was a link to the internal Veeam forums for our system engineers. Removed that link from the post.

To troubleshoot Hotadd/Direct SAN issues, please contact our technical team for logs review, if you haven't done it yet. The log snippet you've posted is not enough to understand where the issue is.

Thanks!