Skip to content

NetApp Storage Shutdown

Graceful shutdown runbook for the JRCAI DM5000H (ONTAP) storage cluster.

Cluster: JRCAI_CLUSTER (JRCAI_CLUSTER-01 / JRCAI_CLUSTER-02, HA pair)

Prerequisites

  • SSH access to the cluster mgmt LIF: 10.22.154.110
  • SP/BMC IPs:
    • JRCAI_CLUSTER-01: 10.22.154.113
    • JRCAI_CLUSTER-02: 10.22.154.114

SP/BMC IPs are only reachable from inside the network

SP ports (.113/.114) are not reachable directly (e.g. from a Mac on the office network) even though the mgmt LIF (.110) is — a known L2/routing gap, likely because SP ports are on a separate physical path. Not yet root-caused, and not blocking: just run the SP console steps from server02 or login01.

Steps

  1. SSH to cluster mgmt

    ssh admin@10.22.154.110

  2. Suppress AutoSupport for the maintenance window

    system node autosupport invoke -node * -type all -message "MAINT=8h Power Maintenance"

  3. Skip the epsilon step — single HA pair, only 2 nodes.

  4. Get SP addresses (already known above, but reconfirm)

    system service-processor show -node * -fields address

  5. From server02 (or login01), open 2 SSH sessions — one to each node's SP — then switch to console:

    ssh admin@10.22.154.113   # -> BMC JRCAI_CLUSTER-01>
system console

    ssh admin@10.22.154.114   # -> BMC JRCAI_CLUSTER-02>
system console

    Exit console back to the BMC prompt later with Ctrl-D or Esc-T, depending on platform.

  6. Back in the main cluster mgmt SSH session, halt both nodes

    system node halt -node JRCAI_CLUSTER-01,JRCAI_CLUSTER-02 -skip-lif-migration-before-shutdown true -ignore-quorum-warnings true -inhibit-takeover true

    Confirm y for each prompt.

    Note

    The mgmt SSH session will disconnect once nodes start halting — this is expected, not an error.

  7. Watch the 2 console sessions on server02 until both show:

    LOADER-A>
LOADER-B>

  8. Once both show LOADER, it's safe to cut power (PSUs or PDU).

Bringing It Back Up

  1. Power on the chassis and let both nodes boot normally.
  2. Verify with cluster show and storage failover show.

  3. End the maintenance AutoSupport suppression:

    system node autosupport invoke -node * -type all -message "MAINT=end"