HP MSA2042 Help

[Jadwiseman]

Hi all,

I need some help working out what I am doing wrong here, I'm hoping that I am just missing something obvious, but for the life of me my brain is a little bit too frazzled at the moment and if anyone could give me a hand it would be massively appreciated.

The goal: Use Veeam Backup & Replication to backup VMs located on two seperate vSphere 6 farms with dedicated storage over fibre channel.

vSphere Farm 1 is 5x HP DL380 G9 servers, all with HPE 82Q 8Gb Dual Port PC HBA's. They connect to a HP 8/8 Base 8-port enabled B-Series SAN switch. This then connects to the storage which is a HP MSA 2042 with 8Gb SW B-Series SFP+

vSphere Farm 2 is 3x HP DL380 G8 servers all with HPE 82Q 8Gb Dual Port PC HBA's. They connect directly to a HP MSA 2040 over FC, again with 8Gb SFP's.

The Veeam server is a physical HPDL380 G9 server, running Windows Server 2012R2 with two seperate HPE 82Q 8Gb Dual-Port FC HBA's installed. Two ports on one of the FC HBA's connects directly to the MSA 2040 for Farm 2. The other FC HBA 2 ports connect into the B-Series SAN switch.

The problem - the Veeam is backing up Farm 2 perfectly without issue, displaying [san] as direct access. Farm two however is failing with direct SAN access, and is falling back on [ndb], which is obviously slower.

I have been on to the MSA 2042 in Farm 1 and made sure the initiators are mapped to the datastores correctly just like on the other cluster.

I have been wondering if it is something to do with MPIO on the Veeam server not being setup correctly, however I have followed the HP guides, so I am unsure.

Is anyone able to spot what is wrong here... i've scoured HP's documentation library for days, reading about 15 different PDF's trying to understand this, but I am failing to see the issue.

Again any help would be greatly appreciated.

[Andreas Neufert]

I would say "it depends" on the configuration you had done.

Look at the disk manager and device manager, do you "see" the voumes from the MSA2042 (connected through the FC switch)?
If no, check SAN Zoning on the switches and LUN Masking configuration on the MSA.

If you see the disks on the windows System (marked as "offline" is OK).
Go in Veeam to Backup Infrastructure - Managed Servers and select the Backup Server with the disks attached and run rescann (right click on the server to see it).

Run the job again.

Please report back if one of these solved it. If not we can discuss additional options.

[Jadwiseman]

Hi,

Firstly thank you for replying!

The disks are appearing offline correctly on the Veeam server for the two mapped LUNs in farm 1 - I have also run rescan on the backup infastructure, but this unfortunately is not working either and the jobs run as [nbd]

The local backup proxy is set to 'Direct storage acces', and works perfectly for farm 2. If disable the 'Failover to network mode' option, and re-run the backup job it fails with the following:

"Unable to allocate processing resources. Error: no backup proxy is able to process this VM due to proxy processing mode restrictions"

[Andreas Neufert]

Then I have no idea. Please open a support case.

Mabe just for a test. Create a new volume on the MSA2042 (10GB) and mapp it to VMware and Veeam (rescann). Then create a new VM with a 40MB thin disk on it without to install anything.
Create a new backup job and back it up. Just for a test.

[blithespirit]

question, why have you got two FC ports direct and one through a switched (I assume fabric), redundancy is fair enough but I would be checking my permissions on those fabrics and your zones.

Create a LUN and present it through the same zone as your farm 1 link to VBR, then try to mount and online the volume. Does this work. I have a feeling it's not MPIO but your zoning and target/initiator permissions.

Can start there and cross things off the list

[Jadwiseman]

Hi,

Thank you for getting back to me. I have created a new LUN with the same permissions on Farm 1 MSA 2042, and presented it to the Veeam server via the FC fabric switch. This appeared correctly in disk manager - I was able to make it online, initialize it and then write data to it - so unfortunately I do not think this is a permissions/zoning issue as this has the exact same setup as the other two LUN's presented to the Veeam server that holds the VM's that are being backed up.

Farm 1 goes via the FC switch as we have 5 ESXi hosts in that Cluster - The fibre B-Series switch comes with 8-ports enabled, so that allows us 2 connections to the storage MSA 2042 controllers, 5 for the hosts, and 1 for the Veeam server using up all 8 ports. Farm 2 is located in a seperate rack in our datacenter and with that cluster only having 3 hosts, we can direct attach to the MSA 2040 using 3 of its 4 FC ports for hosts, and 1 for the Veeam server.

I was going to try attaching the Veeam server FC ports directly to the MSA 2042 in farm 1, however the SFP's I have are not compatible with the MSA - so I would need to buy a couple of new MSA SFPs just to test this.

[Jadwiseman]

Hi,

This problem has now been resolved!

It was luckily my own fault that this issue was occuring, and not a bigger problem. The problem being that the backup proxie 'Connected datastores' field was set to 'Manual selection' - this included the datastores for farm 2, and not farm 1 meaning that it was never going to backup using [san] mode.

I am glad this was an easy fix in the end - Veeam support were excellent .

[Andreas Neufert]

Thanks for the feedback.