Skip to content

REQ 132

sergio velderrain edited this page Nov 12, 2026 · 1 revision

REQ-132: Preserve an operator-approved storage reserve

Requirement

The native trial planner shall refuse artifact staging or VM dispatch when the approved internal-NVMe free-space reserve cannot be preserved.

Rationale

The imported Windows disk, vendor cache, and NSIS payload can consume tens of gigabytes. Reclaiming space by deleting VMs, models, ISOs, caches, or operator files without target-specific approval would turn a capacity problem into data loss.

Fit Criterion

Before a disk-growing action or one-use provisioning claim, the planner compares current free bytes with the pinned image's expected import high-water, artifact-cache budget, and configured retained reserve. Insufficient space returns the available and required byte counts plus a redacted, read-only inventory of candidate paths/sizes and asks the operator where to reclaim space. The planner never deletes, prunes, moves, compacts, or overwrites operator data and never consumes the VM dispatch.

Verification

TEST-133 exercises below/equal/above-reserve cases and verifies no permit claim, VM inventory change, or filesystem deletion occurs when the reserve would be breached.

Source

SYS-REQ-034; REQ-128; epic #250; P2 #255

Priority

High

TestID

TEST-133

TestArtifact

Planned synthetic free-space fixtures and a local NVMe preflight record with exact available/required/reserve byte values, unchanged ledger, inventory, and operator paths.

CodeRef

Planned caller-side capacity estimator around scripts/assisted_benchmark_virtualbox.py and the run-scoped vendor cache. No cleanup authority is added.

Status

Planned; current implementation checks a configured threshold but does not estimate full image/cache high-water or produce operator-directed reclamation findings.

Clone this wiki locally