ThinkSystem DM High Availability: Takeover and Giveback

For ThinkSystem DM nodes, an HA pair uses matched local and partner controllers connected to the partner’s storage. Takeover transfers control of partner storage to the surviving node; giveback returns it after recovery.

Events described in the source

• A panic or unexpected reboot can trigger partner takeover, followed by automatic giveback when the failed node recovers and the applicable giveback conditions are met.
• A node failure that prevents reboot, including loss of power to that controller, can cause the partner to continue serving its data.
• Lost partner heartbeat messages can signal a hardware, software or interconnect problem, even without a panic.
• Halting a node without the source’s -f or -inhibit-takeover true override can trigger takeover.
• A reboot without -inhibit-takeover true can trigger takeover. The source calls this the default -onboot behavior; current ONTAP command documentation names the corresponding storage-failover setting -onreboot.
• The service processor can request hardware-assisted takeover after detecting partner failure, if that feature is enabled.

Manual takeover uses storage failover takeover. This is a maintenance operation and is separate from the automatic triggers.

Exceptions and limits

• If power is also lost to the affected node’s storage, ordinary partner takeover cannot restore access to that unavailable storage.
• In a two-node cluster with cluster HA enabled, inhibiting takeover while halting/rebooting a node can stop data service on both nodes. The source discusses cluster-HA and epsilon preparation; follow the matching two-node maintenance procedure before changing quorum settings.

This is an overview, not a complete takeover or quorum-change runbook. Lenovo’s matching English HA overview provides the source terminology and conditions.