Question About RAID and NRAID

[SWDGuy]

Hi, I just setup a 2 node cluster HP P4500 G2
but I´m not sure about the config.
Each Node is on Raid 5 but the Volumes are on NRAID-0. In the case of a failure disk on each node are the Volumes stay online or the sistem will goes down ?

I´m a little confused about it.
I have to put the Volumes in NRAID 0 cause the space the customer needs.

[SWDGuy]

Another questions I have are here.

If I have A cluster with 2 Nodes in Raid 5 and all the volumes are in Nraid 0 I guess that if one of the nodes goes down I going to loose all the data right ?

talking about Patches How do i Know wich patches do i have to install if i have SANIQ 8.5, Do I have to install all the patches found at the HP web site ??

It is necessary to install always the MPIO DSM ??

To tramitate the License for the P4500 I have to use the Mac of Nic Number 1 of each node right?

[Sven Duthoo]

Hi,

The way I see it, if one disk fails in a node, the node is still working. If a disk fails in the other node, that node is also still ok.
So imho your NRAID0 volumes will still be available.
If you lose 2 disks in one node, the node can go down and your NRAID0 volumes will be down as well.

I don't like the idea of NRAID0, but as you already mentioned, the customer needs the space and the customer is always right :-)

Kind regards,

Sven

[Steven Clementi]

Just think of the relationship(s) regarding RAID.


You have raid 5 on your nodes... As stated, you can lose 1 disk and be ok. (each node)

You have Network Raid 0 on your volume... same as raid 0.. if you lose a disk, you are SoL. Well in thise case... the "disk" is actually the whole server so whatever it takes to "lose" the server.. 2 physical disks in the same array (assuming you have 12 disks, you probably have 2 RAID5 arrays.

If so, than theoretically you can lose 1 disk from each array and still be ok, but losing a 3rd disk would drop the box.


Steven

[SWDGuy]

Thanks, for the info.
Now Im having problems doing the failover test.
when I disconnect one of the ports of the Node the copy process fails so I guess I have to bond the Nics and check the iSCSI initiator configs

Any clue ??

[Steven Clementi]

If you did not bond the nics to begin with... than yes. The default is not bonded.


Steven