StoreVirtual Storage
Showing results for 
Search instead for 
Did you mean: 

Adding 2 P4300 nodes to an existing cluster


Adding 2 P4300 nodes to an existing cluster

Imagine a VMware vSphere infrastructure based on 2 ESXi hosts and a P4000 cluster based on 2 P4300 nodes.
The P4000 cluster hosts 2 LUNs that are accessed by the ESXi nodes via iSCSI.
There are 2 iSCSI initiators (VMware Software iSCSI adapters) on each ESXi node, so each LUN on the P4000 server is connected to 4 (vMware) Software iSCSI adapters.
All LAN connections are redundant, so each ESXi host has 2 LAN interfaces dedicated to iSCSI to connect to the P4300 nodes thru 2 switches and a dedicated VLAN.
Everything works fine, now.

I need to add two new P4300 nodes.

After reading the available documentation and looking at the forums my idea is connecting them to the existing iSCSI dedicated VLAN, adding them to the existing P4000 management group, adding them to the existing cluster.
At that point I presume I will be able to enlarge the existing LUNs.
After rescanning all the storage, vSphere should be able to enlarge the existing VMFS volumes.
In my opinion (am I correct?) I should not lose anything of the LUNs contents.
There is a Failover Manager virtual appliance but I don't think I should do anything to change the configuration.


To be honest it looks too much simple…
Am I forgetting any relevant step?
Is there anything else I should consider?




Honored Contributor

Re: Adding 2 P4300 nodes to an existing cluster

its that simple.


if you want you can do a rain dance for good luck before you join the new nodes to the management group :)


its really simple and easy.


Only suggestion I have is that you add the two nodes to the mgt group and then use the EDIT CLUSTER option to add BOTH NODES at the same time because it will force a restripe of your system and that take a long time if you are actually almost full on your current two nodes.


You will also want to keep an eye on your management group local bandwidth setting.  The default is 16MB/s, but if you have too much load on your system latency will increase on your production environment.  You can fix that by turning that value down, but that will increase the time it take to do the restripe....  not a big deal unless you are impatient since this happens on the live system and doesn't stop production.


Before you ask...  the time it will take for the restripe will depend on your data and system load... it won't happen faster than what you set at the management group level for MB/s rebuild rate, but it can go slower.



Occasional Visitor

Re: Adding 2 P4300 nodes to an existing cluster

I have just purchased two new P4300 (8Tb SAS) nodes like above. I already have two P4300s (8Tb SAS) as a single site using Network RAID 10 for resillience plus a FOM running in another site. I have two x3800's connected running as Windows 2008R2 cluster servics for File and Print Sharing. 


As I'm looking to expand the storage capacity of the existing configuration I'd like to know how the P4300 will arrange the data if I add the two nodes to the existing site.


Will it give me 4 nodes holding the same data but not increase the capacity or will it grow the storage by 8Tb (RAW) and sort out all the re-striping without a problem?


Ultimatley I'd like to have two nodes with 16Tb RAW being replicated (RAID10) to the other two 16Tb RAW storage nodes?


Honored Contributor

Re: Adding 2 P4300 nodes to an existing cluster

since  your new nodes are the same as your current nodes if you add the new nodes to your cluster your total usable space will double.


When you add the new nodes the cluster will initiate a data restripe for each LUN.  Once completed, your data will be mirrored according to your set protection level.  Assuming you keep it NR10 and you have four nodes, the data will go as follows:


Node1 -data "AB"

Node2 -data "AB"

Node3 -data "CD"

Node4 -data "CD"


This way, you can lose node 1 AND 3 or 2 AND 4 and not lose availability.


Just keep in mind that when you add the two nodes, you should add them both at the same time so you don't have to restripe your data twice and also that the order of the nodes is what you see in CMC if you expand your nodes in the management group.