ONTAP Discussions

cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 

ONTAP Discussions

Ask a Question

Convert FAS system to shelf DS4246

potapius
‎2018-01-24 06:13 AM
9,741 Views

Hi!

need help for planning change storage system from FAS2554 to FAS8200 without losing data (minimal).

 

My customer have FAS2554 HA system (CDOT 8.3.2) with following configuration:
Expansion: 3xDS4246(Enc ID:01,02,03) and 1xDS2246 (Enc ID:11)
Stack: only 1 (ID:00 controllers chassis, ID:01 first exp, ID:02 second exp, ID:03 third exp and ID:11 fourth exp)
ADP partition: 2 root aggregate on internal hdd (Enclosure ID:00); ag0-nod1(10 partitions for root) ag0-nod2 (10 partitions for root). Also, on P1 partitions created data aggregate aggr1-n1 and aggr1-n2.
Aditional data aggregate: 4 in aggregates for node1 and 4 in for node2. All with 70-80% data capacity used.
Spare HDD: 8in for all system

 

The new system FAS8200 shiped with CDOT 9.2 and not including additional expansion with HDD. Also ordered "DS4246 Kit FAS2240-4 FAS2554" which will be used for conversion FAS2554 to shelf.

 

  1. In the following schema - I see next steps:
    FAS8200 system not support ADP. Then required create new aggregate in RAID-DP type on 4HDD on each controller for new root, then migrate root volumes to this aggregate (without downtime).
  2. move data volumes from data aggregate aggr1-n1, aggr1-n2( data partition hdd) to other data aggregates system with free space (if not enough - migrating to different storage, IBM for example). After complete - delete all partitions aggregates and zeroing hdd.
  3. shutdown FAS2554 controllers and convert this chassis to DS4246, using DS4246 Kit FAS2240-4 FAS2554 R6 (X897A-R6)
  4. power up exp ID:11 and change id from 11 to 04. Recycle power for accept change.
  5. re-cabling converting shelf to last stack shelf, then power up this and change Enclosure ID from 00 to 05. Recycle power for accept change.
  6. connect all stack (ID:01,02,03,04,05) to new FAS8200 and power on in maintenance mode.
  7. setup new owner ID for root aggregate and boot_ontap on each controller.
  8. Running diagnostics on the chassis. Checking task.

 

I write this plan but I could not find exact answers to several questions:

Сhange shelf ID does not affect the available aggregates and volumes with data on it?

FAS8200 system ccan booting with vol0 (cluster root volume)if CDOT versions of the old and the new system are different? Or before replace chassis I need upgrade FAS2554 to 9.2?
FAS8200 system can booting with vol0 (cluster root volume)if hi lives on partition aggregate?

Preview file
51 KB
0 Kudos
View By:
Tags (1)
6 REPLIES 6

aborzenkov
‎2018-01-24 06:38 AM
9,732 Views
wrote:

The new system FAS8200 shiped with CDOT 9.2

...

FAS8200 system not support ADP.

Wrong. 9.2 supports ADP for 8200 and you can even order 8200 with ADP from factory.

 

As for other questions, it is not clear what root volumes you mean. It sounds like you want to migrate 2554 root volumes to 8200 which will not work and is not needed. Just add 8200 to the same cluster, migrate volumes and LIFs to it and remove 2554 from cluster.

0 Kudos

potapius
‎2018-01-29 12:13 AM
In response to aborzenkov
9,640 Views

Thenk for you answer!

 

The problem is that the FAS8200 was bought without disks, I can not expand an existing cluster (because not have cluster switch) or create a new cluster (because not have additional HDD) for data migration.

 

You write, that root volumes from the FAS2550 system will not work on the new FAS8200 system. Is this related to the CDOT version, or because different controller models?

 

I understand correctly, replacement of controllers on a cluster requires a cluster re-creation?

0 Kudos

aborzenkov
‎2018-01-29 01:35 AM
In response to potapius
9,625 Views
wrote:

The problem is that the FAS8200 was bought without disks, I can not expand an existing cluster (because not have cluster switch) or create a new cluster (because not have additional HDD) for data migration.

 

You write, that root volumes from the FAS2550 system will not work on the new FAS8200 system. Is this related to the CDOT version, or because different controller models?

 

I understand correctly, replacement of controllers on a cluster requires a cluster re-creation?

OK, sorry for confusion, I misunderstood your case. As far as I can tell it should work. You need to follow Controller Hardware Upgrade Express Guide, Upgrading controller hardware by moving storage chapter. If possible, consider arranging for NetApp or partner professional service to assist you.

0 Kudos

potapius
‎2018-01-29 03:59 AM
In response to aborzenkov
9,613 Views

Thank you for documentation link. But in this doc I not find answer for my question:

 

  • FAS8200 system ccan booting with root aggr, if CDOT versions of the old and the new system are different? If yes, than:
  1. migration root volume from adp partition to normal RAID group does not need, because  FAS8200 with ver 9.2 support ADP
  2. after completing "Preparing for an upgrade when moving Storage" (page 13-16) and "Installing new nodes" (page 16) next steps "Setting up the new nodes" I must perform step 6 - "Install new software first". But why boot from the HTTP server, if the controller has a bootable card with the 9.2
  3. Section "Reassigning disks to the new nodes" reguire destroy root aggr and reassign hdd to new owner ID. After accepting this change the system itself will create a configuration backup? How it will estore configuration information from the root volume to the boot devices, if I destroy root aggr?
0 Kudos

aborzenkov
‎2018-01-29 04:19 AM
In response to potapius
9,555 Views
wrote:

FAS8200 system ccan booting with root aggr, if CDOT versions of the old and the new system are different?

It always makes sense to use the same versions on old and new controllers when you do head swap; it minimizes risk of something going wrong.

wrote:

Section "Reassigning disks to the new nodes" reguire destroy root aggr and reassign hdd to new owner ID. After accepting this change the system itself will create a configuration backup? How it will estore configuration information from the root volume to the boot devices, if I destroy root aggr?

It talks about new controller root aggregates, not old controller.

wrote:

after completing "Preparing for an upgrade when moving Storage" (page 13-16) and "Installing new nodes" (page 16) next steps "Setting up the new nodes" I must perform step 6 - "Install new software first". But why boot from the HTTP server, if the controller has a bootable card with the 9.2

One reason is to make sure you have exactly the same versions on both old and new controller. Another reason is to wipe clean boot media to avoid accidental boot with wrong configuration.

 

I once more suggest to find someone with good NetApp experience or at least make sure to secure NetApp support during this activity. With 7-Mode it was possible to fix almost any mistake during head swap. I am not as sure with C-Mode.

 

 

0 Kudos

AlexDawson
A-Team Tech Advisor A-Team Tech Advisor
‎2018-01-30 09:24 PM
In response to aborzenkov
9,536 Views

Headswap on cDOT is a little more complex, in that the controllers do have some boot data stored on them, but it can be reloaded from disk in case of a headswap. As identified, upgrade to a matching version would be the recommended approach here, then swap embedded controllers for IOM6/12s, then boot to maintenance, then reassign disks, then load backup config from flash.

 

But the key thing is what is recommended twice - I also recommend you engage NetApp support or professional services from a PS partner (or NetApp), but we do have documentation available outlining the process of upgrading by "moving storage" available in this document - https://library.netapp.com/ecm/ecm_download_file/ECMLP2540637

0 Kudos
All Community Forums
Public