#Planning to move the 2 node AFF 400 Netapp from one datacenter to another.

1 messages · Page 1 of 1 (latest)

shell imp
#

I did search the netapp site and see how to gracefully shutdown the array documentation as below but looking for any one who did and share the feedback.
https://kb.netapp.com/onprem/ontap/hardware/How_to_perform_graceful_shutdown_and_power_up_of_all_ONTAP_nodes_in_a_cluster

NetApp Knowledge Base
How to perform graceful shutdown and power up of all ONTAP nodes in...

Events such as scheduled site power outage, physical system move and data center wide maintenance may require a graceful shutdown of ONTAP equipment. This KB details the prior to shutdown best …

misty vessel
#

Done it a bunch of times. (when I was a customer and Partner PS) It's pretty straight forward.

  1. Label all your cables and shelves.
  2. Stop IO to the system.
  3. halt cluster.
  4. power down cluster (chassis first)
    *note: If you have a DS460 you will need to remove the drives from the shelf chassis. *
  5. Move it.
  6. cable it back up.
    7 Power on the shelves. (wait about a minute)
  7. power on the controllers.
  8. check cluster health and HA afterwards.
shell imp
#

Thank you for the reply @misty vessel . I have 3 questions from above steps.

  1. On our 2 node cluster we have only CIFS and NFS and i thought for stopping IO to down the CIFS server which basically stops all IO's. Is there any command for NFS to stop IO? If there is better way to stop IO for both CIFS and NFS, please let me know.
  2. Even though we say to CIFS users to save the work what ever they do , some users definitely don't do that. Is there a clean way to disconnect all cifs client sessions before i shutdown the cluster.
  3. I thought to halt both the nodes but i am not sure on step 4. (power down cluster). Can you please explain a bit.
misty vessel
#

  1. There is vserver nfs and cifs stop . Typically though servers/hosts etc get shut down first. So you're just left with a few CIFS stragglers. (but this can vary from enviroment to enviroment)
    https://docs.netapp.com/us-en/ontap-cli-98/vserver-nfs-stop.html

  2. outside of stopping/terminating cifs i'm not aware of any. I recall shutting down systems before and open cifs sessions reconnected after the move. I don't think i've ever seen a shut down with all CIFS gone, unless networking shut off access to the data center so clients couldn't access it.

  3. You do the whole cluster at once. You can connect to it via SP or Serial connection.

Here's the command from the KB -

system node halt -node * -skip-lif-migration-before-shutdown true -ignore-quorum-warnings true -inhibit-takeover true

it will halt each node, while not moving lifs, it gnores quorum warnings, and no takeovers.

#

I would also pause any snapmirror ops as well.