Skip to main content
Cloud Volumes ONTAP

Prevent disruptions during Azure maintenance for Cloud Volumes ONTAP HA pairs

Contributors netapp-manini
PDFs

You can configure Cloud Volumes ONTAP HA pairs to suppress automatic takeover and giveback during Azure scheduled maintenance. This optional feature is available with Cloud Volumes ONTAP 9.19.1 and later and is disabled by default.

Note This feature is available as a public preview with Cloud Volumes ONTAP 9.19.1.

Microsoft Azure can briefly pause a VM during scheduled maintenance. In a Cloud Volumes ONTAP HA configuration, ONTAP might treat the pause as a node outage and move storage operations to the partner node (takeover). ONTAP then automatically gives the operations back when the paused VM resumes (giveback). This behavior can disrupt active client connections. If you enable this feature, the HA pair continues operating through an Azure scheduled-maintenance freeze without automatically initiating takeover and giveback.

About this task

  • Client I/O might stall during an Azure VM freeze. ONTAP does not automatically initiate takeover, so client I/O remains on the affected node rather than moving to the HA partner. I/O might stall until the node resumes.

  • The duration of an interruption depends on the Azure VM freeze. A longer freeze can result in a longer I/O interruption.

  • Normal HA protection remains active for other failure conditions. The feature suppresses takeover and giveback only during Azure scheduled-maintenance freezes. Other conditions that normally trigger takeover, such as a node failure or panic, continue to do so.

  • The feature applies to both nodes in an HA pair. The maintenance-event behavior is configured on both nodes and cannot be enabled on only one node.

  • The feature persists across reboots and remains in effect until you disable it. Disabling it restores the default scheduled-maintenance takeover and giveback behavior.

Before you begin

  • Your Cloud Volumes ONTAP system must run 9.19.1 or later.

  • You must have Cloud Volumes ONTAP administrator access to complete this procedure.

  • Verify that your workloads can tolerate temporary I/O stalls during an Azure scheduled-maintenance freeze and that continuing through the freeze is preferable to the default takeover and giveback response.

Prevent automatic takeover and giveback during Azure scheduled maintenance

Steps

  1. Identify the system ID for the Azure HA pair. For information about identifying resources and making API calls, see Your first API call.

  2. Send a PUT request to the following endpoint, replacing {id} with the system ID:

    PUT /occm/api/azure/ha/working-environments/{id}/opt-off-maintenance-event

    Set enabled to true in the request body:

    {
  "enabled": true
}
Result

A successful request returns HTTP 202 with an empty body.

Important The API updates the nodes in the HA pair sequentially. If a request fails before the operation completes, verify the setting on both nodes before retrying the request because the nodes might have different settings.

Disable the feature to restore default behavior

To restore the default takeover and giveback behavior during Azure scheduled maintenance, disable the feature.

Steps

  1. Send a PUT request to the same endpoint, replacing {id} with the system ID:

    PUT /occm/api/azure/ha/working-environments/{id}/opt-off-maintenance-event

    Set enabled to false in the request body:

    {
  "enabled": false
}