Storage Boards Cleanup
To make it easier to find information about HPE Storage products and solutions, we are doing spring cleaning. This includes consolidation of some older boards, and a simpler structure that more accurately reflects how people use HPE Storage.
HPE StoreVirtual Storage / LeftHand
cancel
Showing results for 
Search instead for 
Did you mean: 

Adding an additional HP Storevirtual, cluster question

bruun
Occasional Contributor

Adding an additional HP Storevirtual, cluster question

Currently there's a single HP Storevirtual running, a cluster is already configured and the iSCSI (Hyper-v) are connecting to the clusters VIP. The plan is to add a additional HP Storevirtual and switch for redundancy, the question is: can i add the additional HP Storevirtual and add it to a cluster, configure network raid 10 and don't loose any data in the progress? To me, it seems to be possible because it's mirroring. Unfortunately i can't find any documentation that describes the process.

Also going to connect a new switch. At the moment all 4 interfaces are bound into a ALB and connected to a single switch. I assume i need to edit the ALB and remove 2 interfaces from this group. After that create a new group and add the 2 interfaces. The result is 2 bound groups.

I think these are the steps:

1. Install and configure the new switch
2. Remove 2 interface from the ALB group
3. Create a new ALB group and add the 2 interfaces
4. Connect the two interfaces to the new switch
5. Install Fail over manager virtual appliance
6. Install and connect the new HP StoreVirtual and add it to the cluster
7. Install the latest firmware, on both HP StoreVirtual
10. Create a raid 10 network (Data will be kept in place?)

I think the steps above should do it. 

 

3 REPLIES
Stor_Mort
HPE Pro

Re: Adding an additional HP Storevirtual, cluster question

Hi bruun, 

Firstly, congratulations on  moving to a fully redundant configuration! StoreVirtual clusters can give you very high reliability at a very reasonable cost. If you are careful, you may be able to add the new node without any downtime.

There are three main tasks that must be done to reconfigure your cluster for 2 nodes. I would suggest doing them in this order.

1. Configure network, bring up the new node in CMC, add it to the management group.

Install the new switch. There must be a link between the switches to ensure full redundancy in some failure situations. It is not necessary to create two ALB bonds - in most cases you will connect two storage ports to switch A, two storage ports to switch B, and use the ALB bond to make the four links work as one link with one IP address. Wire the second node the same way - two ports to switch A, two ports to switch B. Your servers should also be connected to both switches.

Find the new node in the CMC. Make one bond with the four NICs. Ping each node and server from each other to verify connectivity. Resolve any switch configuration issues before moving on - a stable, low latency, zero packet loss network is necessary for best results with iSCSI.

Add the new node to the management group.

2. Start manager and add Fail Over Manager or Quorum Witness file to management group.

Start the manager on the new node. Install a Fail Over Manager on a host that is not dependent on storage provided by this SAN. (Download FOM from www.hpe.com/info/storevirtualdownloads) Alternately, you can use an NFS file as a Quorum Witness. The NFS server must be independent of the SAN. Managers, FOM and Quorum Witness are described in chapter 9 of the HP StoreVirtual Storage User Guide, available at the download site I just mentioned.

3. Add the new node to the cluster.

Look in chapter 10 under "Adding a storage system to a cluster" for the steps. Your existing  Network RAID-0 volumes will be restriped automatically to Network RAID-10 without going offline. This may take several hours depending on how much data is being restriped.

It is worth noting that adding another node doubles your cluster storage space, but converting the volumes to Network RAID-10 uses twice as much space as before. For instance, if you were using 40% of the cluster space before, you will still be using 40% of the cluster space after adding the new node and making the volumes redundant with NR10.

If everything is working as expected, you can apply patches and updates with the CMC after the restripe has completed.

I am an HPE employee - HPE StoreVirtual Support
bruun
Occasional Contributor

Re: Adding an additional HP Storevirtual, cluster question

Thanks for the detailed answer!

Got one more question: for the 2 nodes i'll have to install the failover manager, the failover manager needs to communicate with the VirtualStores. The Hyper-v server is a HP Proliant DL 380 and has 4 NICs:
2x for MPIO
2x teamed
These are connected to a HP Procurve switch. There will be 1 vlan for the production and one for iSCI.
The problem is that the VM running on top is using the team that's connect to the production VLAN, and by default it's unable to communicate with the StoreVirtuals in the iSCSI vlan. What's the best way to solve this? I've attached a picture, for simplicity there's one switch in the drawing (there will be 2 in production)

Stor_Mort
HPE Pro

Re: Adding an additional HP Storevirtual, cluster question

The FOM VM must be connected to the iSCSI network so it can communicate with the SAN nodes. In the VM Settings, select the correct network for the virtual NIC. Then open a Console window for the FOM and change the IP Settings so it has an IP address on the iSCSI network. You should be able to find it with the CMC and add it to the SAN Management Group.

Don't put the FOM on a SAN volume. Be sure to put the FOM on local storage on the Hyper-V host so that it can keep running if there is a SAN issue.

I am an HPE employee - HPE StoreVirtual Support