Repairing a Node in a P4300 SAN cluster

[mdae]

Hi,

I have a running P4300 "Starter San" 2 nodes running with network raid used as storage for two XEN host servers

One of the machines in the cluster is giving a E01020207 EID_BBU_FAULTY error. 

I know it's probably the battery so we will have to replace that.

But how do I proceed?

I can A:  turn off all XEN VM's then do a "Shut down management group" and it will shutdown the SAN properly

replace the battery and start the SAN and the XEN HOSTS.

While I assume this works I wonder if I can do this without downtime.. but how? 

I'm afraid that if I will shutdown one of the SAN nodes and turning it on again it will sync back the "old" data from the repaired node.

Can someone point me in the right direction?

[Dirk Trilsbeek]

if you have all your volumes running in Network Raid 10 (mirrored) and if you have a failover manager in your management group, you can just shut down the faulty node, repair it and reboot the node without the need for a downtime.

[mdae]

Thank you for the reply ..

I don't have a failover manager since there isn't one available for XEN (only for esxi)

and we don't have the rackspace to place a physical server.

Is there a other way ?

[oikjn]

you really need to get a FOM installed SOMEWHERE...  you don't have a single windows 2008 server running?  There is a hyper-v FOM too. 

Right now, without a FOM, you have no redundancy with your setup and if either system goes down you lose access to everything (you won't lose data, just availability).

There is a feature in CMC that you can turn on a "virtual manager" on for the node you are going to keep up and that should give that node the equivalent of two votes for quorum.  You can find that virtual manager by right clicking on the managemetn group name in CMC.

You need to push your company to get a FOM installed.  Spend 300$ and get a 1U supermicro atom based server and install hyer-v server or esxi on it and load up that FOM.

[mdae]

Could this also be placed on one of the XEN hosts? (local storage) (hyper-v machines convert nicely to xen)

In the meanwhile I'll try to convince them to get another server.

And thank you for pointing out  the virtual manager.

[oikjn]

I haven't tried, but I think it might work.  I do know they use the Hyper-v integration services for things like controlled shutdown when the host reboots, but it would be 1000% better than using no FOM. Might wanna ask support to be sure, but I would be willing to bet that unless you pushed the issue with them they will just say "unsupported".

[oikjn]

the FOM really requires almost no resources so I think even something as dinky as an atom based server can fit the bill... but I haven't tried it so don't quote me on that :)

[Emilo]

Hello,

for the simple operation you want to do right now just install the virtual manager, start on it on the node that you will not bring down and you will maintain quorum. After the repair stop the VM. The VM is manual , where as the FOM is automated.

Just install the vmware player on any machince that is on the same vlan, you could even use the machine that is hosting the CMC he player is free and you install the FOM on it.

[mdae]

First of all thank you all for helping me out.

I could convince our client to add a FOM and we went for a Hyper-V FOM and it has been added to the Management Group.

Initially I wasn't able to add the FOM to the management group because a virtual manager was running.

But after deleting the Virtual manager I could add the FOM.

If I need to reboot the FOM (due to patches for the hyper-v host) is there anything I need to do before the reboot?  (of course I wouldn't reboot the fom if something is wrong with the cluster <g>)

Next would be replacing the BBU battery.

So  can I just shutdown the Storage System (Right click storage system and select "Power Off or Reboot")

Or do I have to remove the system from the cluster first?

[Emilo]

Just power off.

Make sure you check the "availablity' tab first

It should be blank