HPE Storage Users Group

Hi,

I have a 3PAR 7400 running OS 3.1.3. It's been out of support for a while and still has some production data on it. It has 16 cages, 8 per node pair. I'm currently evacuating all drives in cages 8-15 that are using loops 3:0:X and 2:0:X.

I'm confident that I can remove the drives successfully (I kept a T400 running by dismissing failed disks. You know...the ones that didn't officially have a problem ) and I'm fairly confident I can unplug each cage and via servicecage and admithw remove cages 9-15.

I'm not sure on the correct process to remove cage 8 containing nodes 2 and 3. After shutting down the nodes and running admithw will the array become a 7200?

I then need to move nodes 2 and 3 to start a new array which I assume I could do by running an ootb so they become nodes 0 and 1. Split the license and Roberts your mothers brother?

Any help or pointers would be appreciated!

Good luck.

If you are able to do this, you'll get 2x 7400-2n. (7200 nodes have less core enabled, c-versions are equal)

However...
#1 - I'm pretty sure that node 2 and 3 will stay as node 2 and 3 when you split them, which is unsupported from HPE and probably not tested.
#2 - Your license is tied to the serial of the system, so you can't split the licenses.
#3 - The serial is coded on the system so you're most likely unable to change this if trying
#4 - 3PAR WWN is tied to the serial, so if you're unable to change that both system will have the same WWN preventing you from using them on the same SAN.

So I don't think you have have any big problems removing 8 cages and 2 nodes from the system (offline), but being able to use those two nodes for something is a different story.

Maybe look on the second hand market for a licensed 7200/7400 and use the drives there?

_________________
The views and opinions expressed are my own and do not necessarily reflect those of my current or previous employers.

Hi @MammaGutt

That's an interesting point about nodes 2 and 3 staying as 2 and 3. I remember our original T400 only had nodes 2 and 3 until it was extended with a second cab later and gained nodes 0 and 1.

Today I evacuated and dismissed 192 disks. No amount of running tunesys would remove the data. It was only when I copied the CPG and moved the volumes did the data move. I still had somewhere between 0 and 4 chunklets left on each disk that had to be moved manually using movepdtospare.

Tomorrow morning will be spent correctly zoning all hosts to only see nodes 0 and 1.

After I've labelled all cages and nodes I'll be running a servicecage -remove to remove the cages including the cage holding the nodes. I believe that although one cage houses the nodes it's still logically seen as a disk cage which can be serviced.

Then I'll be shutting down nodes 2 and 3 (with a good gap between shutting down each one)

After removing the cross links between the node pairs and removing the network cables I'll boot the nodes, run a ootb and see what state it's in.

I'm a bit stuck at the moment. The servicecage remove cage# command fails as the cage has active ports. With a T400 style unit that has dedicated fc ports to each cage it's easy to shutdown but with the SAS loop I'm a bit lost. I've shutdown all disk ports in one cage by using servicemag offloop cage# # where the last is disk 0-23 but it still wont remove.

I really wish HP would document these processes. Owning a storage device and not been able to fully manage it is a real pain.

Setting ports 2:0:1, 2:0:2, 3:0:1, and 3:0:2 offline via the gui is not the same as running controlport offline 2:0:1. After running the controlport commands I was able to run a servicecage remove cage# to remove each cage.

ALL ports are now offline on nodes 2 and 3. Just waiting for confirmation that the nodes can now be safely powered down.