Nearly 500 GB of allocated-but-empty storage reclaimed on the virtualisation hosts

Thin-provisioned disks only consume space as they are written to, but two of them had been fully allocated by earlier copy operations that wrote every block — including the empty ones. The space was held but held nothing.

The two disks between them were using 644 GB to store 50 GB of real data. Both have been returned to their pools: one by extending a filesystem that had never been grown to match its disk, so the unused region could finally be released; the other by discarding the free blocks directly while the machine was briefly stopped.

One pool went from 46% used to 19%, and from 144% committed to 95% — back under its own capacity. The other dropped from 40% to 14%. Nothing was moved and no data was touched; the space was simply never in use.

This also closes out a long-standing gap: a warning that reclaim had not worked was being masked by a tool that reports the range it attempted to free rather than what was actually freed.