Skip to content

Changelog

All notable changes to S2DCartographer are documented here. For the complete history see the root CHANGELOG.md in the repository.

The format follows Keep a Changelog and this project uses Semantic Versioning.

[1.10.1] — 2026-06-30

Fixed

  • ImportExcel is now an optional runtime dependency. Removed ImportExcel from the manifest RequiredModules list so Test-ModuleManifest and Import-Module succeed in CI without ImportExcel pre-installed. Export-S2DExcelReport loads ImportExcel on demand; other report formats (HTML, Word, PDF, JSON) are completely unaffected.

[1.10.0] — 2026-06-30

Added

  • Graduated health-check scoring engine (#57, AB#267). Get-S2DHealthStatus now adds Weight, MaxPoints, AwardedPoints, ScoreBand, and ScorePercent to every S2DHealthCheck. A CollectedData['HealthScore'] rollup carries the weighted OverallScore (0–100), ScoreStatus (Excellent/Good/Fair/Needs Improvement), TotalAwarded, TotalMax, and the legacy OverallHealth string. All existing output fields are unchanged — fully backward-compatible.

  • Named calculation references via config/health-checks.json (#58, AB#268). All 12 check definitions (weights, thresholds, named bands) externalized to a JSON config file shipped with the module. Thresholds and weights are editable without code changes. Mirrors Ranger's waf-rules.json schema shape.

  • Export-S2DHealthConfig and Import-S2DHealthConfig (#59, AB#269). Export writes the active config to JSON; Import validates and activates an override for the session. Use -Validate for a dry-run schema check; use -Default to reset to shipped defaults. Mirrors Ranger's Export-RangerWafConfig / Import-RangerWafConfig hot-swap pattern.

[1.9.0] — 2026-06-30

Added

  • Provisioning-aware 70% headroom presentation — the 70% amber/warning treatment in the HTML KPI tile, capacity-waterfall planning-line row, and expansion-headroom table now fires only when thin-provisioned volumes are present (a full pool takes thin volumes offline). When all volumes are fixed/thick, the 70% row is rendered as advisory/neutral (blue label, "Advisory — fixed volumes commit footprint up front") because footprint is committed at creation time. The Get-S2DExpansionHeadroom function gains a HasThinVolumes boolean on its output to drive this. Reworded all 70% prose to: "70% is a recommended operational headroom guideline — mainly relevant for thin-provisioned volumes, where a full pool takes thin volumes offline. Not a Microsoft hard limit. The firm storage limits are footprints fitting the pool and the rebuild reserve staying intact."

  • "Size to enter (New-Volume / WAC)" column in all expansion-headroom tables (HTML, Word/PDF). PowerShell New-Volume -Size and WAC parse size suffixes as binary (typing 2TB means 2 TiB). Each headroom row now shows the exact value to type, rendered as {value}TB (e.g. 0.65TB), computed as NewUsableData.TiB rounded down to 2 decimal places so the new volume always fits. Past-line rows show —. A footnote explains the binary-suffix convention. The SizeToEnterTiB property flows through to the JSON snapshot automatically. Identical wording and rounding to Surveyor 2.6.0 so the two tools match exactly.

Fixed / Correction

  • N+1/N+2 maintenance reserve re-framed as compute-advisory — the v1.8.0 implementation incorrectly treated N+1/N+2 as a storage-pool capacity check. Microsoft WAF documents it as a COMPUTE resiliency target (reserve CPU + RAM so a node can be drained for updates or lost without dropping VMs), listed separately from storage. Get-S2DMaintenanceReserveAssessment now returns Status='Info', Meets=$null, and surfaces LargestNodeCapacity as informational context. Check 12 (MaintenanceReserveN1) always emits Status='Pass' at Severity='Info' and never contributes a storage Warning or Fail to the health rollup. HTML and Word report sections re-labelled as compute advisory. All 70% prose corrected to not imply N+1 is a firm storage limit.

[1.8.0] — 2026-06-30

Added

  • Maintenance reserve (N+1) assessment — reports whether the scanned cluster currently has enough free capacity (beyond the rebuild reserve) to drain a node for updates, per Microsoft's Well-Architected guidance. Surfaced as a new MaintenanceReserveN1 health check and a report indicator (Meets / Does not meet, with the required vs available headroom in TB and TiB). Configurable target via -MaintenanceReserveTarget None|N+1|N+2 (default N+1). This is assessment only — it does not change any measured capacity number. On two-node two-way-mirror clusters it notes that each node already holds a full copy, so a drained node keeps data available.

[1.7.1] — 2026-06-30

Added

  • Expansion Headroom report section (HTML, Word, PDF, JSON) — at 70% / 80% / 90% / 100% pool-fill targets, shows the remaining pool footprint and the new usable data you can still store (footprint ÷ resiliency copies), in both TB and TiB, with a Chart.js chart. The 70% row is styled as the recommended planning line. Math is identical to Surveyor's so the two tools reconcile by construction.

Fixed

  • Per-volume capacity efficiency is now derived from the measured FootprintOnPool ÷ Size ratio (ground truth) when available. A two-copy mirror correctly reports 50% efficiency; the previous logic mislabeled it as 25% ("nested two-way mirror") on two-node clusters. Nested two-way mirror, which writes four copies, correctly reports 25%.

[1.7.0] — 2026-06-30

Wave 3 — Validation & report correctness. Locks accuracy against regression and fixes the PDF/Word report rendering defects, plus collection robustness on empty or contended clusters.

Added

  • Cross-tool reconciliation keystone. A byte-valued golden-cluster fixture (tests/maproom/Fixtures/golden-clusters.json), identical to Surveyor's, drives a Pester suite that runs the Cartographer engine and asserts reconciliation within the ±2% gate. 24 Pester cases across 4 clusters. (AB#4633)
  • Concurrent collection guard — a module-scoped flag set at the start of Invoke-S2DCartographer and released in a finally; a second concurrent collection throws a clear terminating error instead of corrupting shared state. (AB#279)
  • Empty-data safeguards — Get-S2DCapacityWaterfall, Get-S2DHealthStatus, and Get-S2DStoragePoolInfo now handle null/empty disk, pool, and volume data gracefully (warn + zeroed result) rather than throwing on an empty cluster. (AB#279)

Fixed

  • PDF capacity-model waterfall bar graph now uses exact byte-derived TB values instead of a truncated 1.0995 approximation. (AB#263)
  • PDF pool allocation reserve boundary line is positioned at poolTotal − reserve (start of the reserve zone) instead of the reserve value, and the pool health bar's x-axis max now includes every segment so healthy clusters no longer clip. (AB#264)
  • Word executive summary table given explicit column widths (35/65), fixed layout, and cell padding so values like 22.44 TiB (24.65 TB) are no longer cramped. (AB#265)
  • Word document overall layout — 1-inch margins, page-width tables with fixed layout, section- header cell margins, and hand-tuned column widths for the wide disk/volume tables. (AB#266)

[1.6.0] — 2026-06-29

Wave 2 — Legibility. Capacity figures are now unambiguous in unit and space. No capacity math changed.

Added

  • The capacity waterfall and HTML report label every figure with both its unit (decimal TB and binary TiB) and its space — each stage is badged FOOTPRINT, DATA (usable), or INFORMATIONAL. (AB#4645)
  • 70% planning line — 70% of available-for-volumes, compared against consumed volume footprint (matches Surveyor's definition exactly). Surfaced as PlanningLine70Pct / IsAbove70PctLine on S2DCapacityWaterfall, as a KPI tile (amber when crossed) and a dashed plan-line on the HTML report chart. (AB#4644)

[1.5.0] — 2026-06-29

Wave 1 — Capacity accuracy. The capacity waterfall is reconciled to the canonical model (docs/capacity-model.md) shared with Azure Local Surveyor.

Changed

  • Resiliency factor is now always derived from each volume's actual NumberOfDataCopies when present. When that value is unavailable, the unsafe default fallback changed from 3.0 to 2.0 (two-way mirror) and the waterfall stage is labeled [ASSUMED — actual NumberOfDataCopies not available] via the new -ResiliencyIsAssumed switch. A real pool value is always preferred over the assumption. (AB#4642)

Confirmed

  • Reserve calculation confirmed canonical — one capacity drive per server, up to 4 drives (raw-byte basis). No change required. (AB#4643)

[1.4.2] — 2026-04-14

Fixed

  • RebuildCapacity health check fired false Critical when pool-member disk deduplication could not assign correct NodeName via StorageNode associations — all disks landed on one node making the check compare pool free against the full pool total instead of one node's capacity. Fix: fall back to pool.TotalSize / nodeCount when NodeName group count is less than nodeCount.

For versions 1.4.1 and earlier, see the root CHANGELOG.md.