Storage Spaces spreads one virtual volume across several physical disks. When a member disk stops answering, the pool does not release it, because removing it at once would break the volume that spans it. The disk stays listed as missing and refuses to be removed until you tell the pool to evacuate its data, which is a different command from the one most people try first.
Storage Spaces groups physical disks into a pool and carves virtual disks out of that pool. Each virtual disk is sliced into slabs that are spread across the members, and redundancy is achieved by keeping more than one copy of a slab.
When a member stops responding, the pool marks it as missing or as lost communication instead of dropping it. Removing the disk immediately would invalidate every slab that lives on it, so the pool refuses, and an unhealthy virtual disk keeps the disk listed even after you have unplugged it.
The state is a safety feature rather than a defect. The disk has to be retired, which orders the pool to copy its data to the other members before the disk is released.
Open the Storage Spaces settings page or the classic Control Panel applet and read the physical drives list. A failed disk shows a warning and a remove link, while a healthy one offers to be prepared for removal.
Retire first. The pool then copies the disk contents to the remaining members, which requires free capacity and, for a two-way mirror, enough healthy disks to end up with a redundant copy of every slab.
Only after the evacuation finishes does removal succeed. If the pool is nearly full, the operation stops partway and the disk remains in the list with a mixture of healthy and missing extents.
Check the virtual disk health after the removal. A two-way mirror needs two good copies of every slab, and if the pool dropped to a single copy the volume reports a warning rather than healthy.
Add a replacement disk that is at least as large as the one you removed, then let the pool rebalance. A smaller disk cannot hold the largest slabs, so the pool leaves capacity unused or refuses the disk.
Verify by reading a sample of files from the virtual volume before you trust it again. A pool that reports healthy but returns errors on old data has already lost a slab, and a backup taken now would preserve the damage.
Because one volume spans several disks, a pool problem is not a single-disk failure and cannot be solved by pulling one drive. Qiling Disk Master can back up the mounted virtual volume to a disk outside the pool, which is the fastest way to protect the data before you change the layout.
Start the program and open Backup. A folder-level File Backup is enough for a data volume, while System Backup is the right choice when the virtual disk hosts Windows itself.
Pick the volume by drive letter and compare its capacity with the size the pool reports. If the volume has no letter, assign one first, because the backup needs a usable path.
Use an external disk or a disk from another pool. Writing the image to a member of the same pool defeats the purpose, because the pool is exactly what is at risk.
Start the task and let it finish without pausing the machine. Then open the image and confirm that a few folders can be browsed; a backup that cannot be read is worse than none, because it creates false confidence.
Because the pool still holds data on it. Retire the disk, or wait until the pool marks its slabs as unrepairable, before removal becomes available.
Sometimes. Recovery software can read the member disk directly and rebuild the virtual layout from its metadata, but only if the disk is still detected by the machine.
No, but some layouts allocate slabs according to the smallest member, so mixing sizes wastes capacity rather than breaking the pool.
The virtual disk status reads healthy and the physical drives list shows no warnings. Read a few files as well before you delete any backup.
For recovering files from a pool volume that has already lost data, see recovering deleted files from a Storage Spaces volume.