Decide whether an S2D server removal also removes its drives

Source facts

Microsoft distinguishes removing a server while retaining its drives in the pool from permanently removing both the server and its drives. When drives remain expected but missing, the procedure does not move their data away; the source describes lost-communication and incomplete-volume states. Permanent scale-in requires sufficient remaining capacity for the volume footprints. The surviving fault-domain count must also support the volume resiliency configuration. Microsoft Learn.

Applicability

State whether the server will return, its drives will move to another server, or its capacity is being retired permanently. Inventory volume footprints and the surviving resiliency layout. Do not equate a quick cluster-node removal with a completed evacuation of the server’s storage.

DSE recommendation

Have the storage owner sign off the intended removal type and the post-change node and drive inventory. For permanent scale-in, calculate remaining capacity and verify the required failure-domain count before scheduling the action. Preserve the planned return or recovery procedure and agree on the acceptable workload impact. Avoid switching between temporary removal and permanent retirement without reviewing the storage consequences.

Verification

Inspect node membership, pool drive membership, and volume health after the approved operation. Compare the observed result with the declared temporary or permanent intent. For retirement, verify the intended data movement and healthy surviving layout before releasing hardware. For temporary removal, track the expected return and unresolved missing-drive state explicitly.

Official references

Microsoft Learn: Removing servers in Storage Spaces Direct. Source reviewed September 8, 2026.