docs: correct stale/false facts in O3 partition plan (TK-11364)

88e393b7fae7d977a88237ad9ef5e96a9661870e · 2026-09-10 07:36:27 -0700 · Steve

The plan claimed the Kamatera volume grow had happened and that it solved the
disk refill. It did not — sda is still 700G with 112d uptime; Steve deferred
Phase 1 on 2026-09-03. A future session reading this would assume Phase-2
headroom that does not exist.

Adds a status-correction banner reconciling the doc against read-only prod
measurement plus the independent homesonspec-size-canary heartbeat:
growth 16 -> 3.24 GB/wk (TK-11125 payload guard), SourceEvidence 55 -> 82GB,
free space 67 -> 44GB (preflight item 1 now FAILS), size canary already armed,
step-1 index drop already applied, and the "retention gets turned ON" pull
trigger permanently dead per Steve's keep-full-history ruling.

Body preserved as written for audit.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JhnABoxWtpF7mbBowoBWhb

Files touched

M ops/o3-partition-and-disk-plan.md

Diff

commit 88e393b7fae7d977a88237ad9ef5e96a9661870e
Author: Steve <steve@designerwallcoverings.com>
Date:   Thu Sep 10 07:36:27 2026 -0700

    docs: correct stale/false facts in O3 partition plan (TK-11364)
    
    The plan claimed the Kamatera volume grow had happened and that it solved the
    disk refill. It did not — sda is still 700G with 112d uptime; Steve deferred
    Phase 1 on 2026-09-03. A future session reading this would assume Phase-2
    headroom that does not exist.
    
    Adds a status-correction banner reconciling the doc against read-only prod
    measurement plus the independent homesonspec-size-canary heartbeat:
    growth 16 -> 3.24 GB/wk (TK-11125 payload guard), SourceEvidence 55 -> 82GB,
    free space 67 -> 44GB (preflight item 1 now FAILS), size canary already armed,
    step-1 index drop already applied, and the "retention gets turned ON" pull
    trigger permanently dead per Steve's keep-full-history ruling.
    
    Body preserved as written for audit.
    
    Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
    Claude-Session: https://claude.ai/code/session_01JhnABoxWtpF7mbBowoBWhb
---
 ops/o3-partition-and-disk-plan.md | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/ops/o3-partition-and-disk-plan.md b/ops/o3-partition-and-disk-plan.md
index 0d6e3b56..4d2e9846 100644
--- a/ops/o3-partition-and-disk-plan.md
+++ b/ops/o3-partition-and-disk-plan.md
@@ -1,5 +1,33 @@
 # TK-10878 — O3 partition + add-disk plan (Steve-chosen path, 2026-08-26)
 
+> ## ⚠️ STATUS CORRECTION 2026-09-10 (claude-run-10878) — READ BEFORE ACTING ON THIS DOC
+>
+> This document was written 2026-08-26 and last revised 2026-09-01. Several of its load-bearing
+> facts are now FALSE. Corrected against read-only prod measurement + the independent
+> `homesonspec-size-canary` heartbeat (2026-09-10T02:31, conn=ssh, verdict PASS):
+>
+> | doc says | actual 2026-09-10 | why it matters |
+> |---|---|---|
+> | "After the disk grew… solved by the disk grow" | **THE DISK GROW NEVER HAPPENED.** `sda` is still 700 G, uptime 112 d, no restart. Steve DEFERRED Phase 1 on 2026-09-03. | ⬅️ **The dangerous one.** Do NOT assume Phase-2 headroom exists. What actually solved the refill was the snapshot offload to Henry + the 09-03 reclaim + the TK-11125 payload guard. |
+> | growth "~16 GB/wk" | **3.24 GB/wk** (size canary, rolling) | TK-11125 payload guard (prod 2026-09-02) cut evidence/SourceRun 4,620 → 384 (–91.7%) at unchanged crawl volume. |
+> | `SourceEvidence` "~55 GB" | **82 GB total / 64 GB heap** | The conversion scratch requirement is correspondingly larger. |
+> | box "~90 % full (~67 GB free)" | **94 % full, 44–45 GB free** | Preflight item 1 (≥75 GB free) FAILS today. Phase 2 is NOT runnable. |
+> | Phase 3A canary "Steve action to arm it" | **already armed and reporting** (conn=ssh, PASS, runway 14 wks) | No action needed; it is the standing guard. |
+> | Sequencing 1 "apply STEP 1 index-drop" | **APPLIED 2026-08-26** (~509 MB, ledgered, undo recorded) | Done, not pending. |
+> | Sequencing 2 "PHASE 1 add disk" | **DEFERRED by Steve 2026-09-03** | Deliberate: fix the growth source rather than feed it. Vindicated — growth is down ~7–10x with zero spend. |
+> | pull-trigger "retention gets turned ON" | **PERMANENTLY DEAD** | Steve ruled 2026-09-02 "KEEP FULL HISTORY — property-history-over-time is the product." Partitioning will NEVER reclaim space here; DETACH-to-cold-archive is the only path. |
+>
+> **Net effect on Phase 2:** still correct architecture, but now justified ONLY on ops/query
+> ergonomics (per-partition vacuum/reindex, cheaper cold-archive, createdAt pruning) — never on
+> size reclaim. Urgency downgraded from "the box is filling" to "the right long-term shape."
+> It remains BLOCKED on the ≥75 GB-free gate. Today's disk pressure is NOT homesonspec
+> (project dir 127 G → 51 G); 64 G of it is `/root/.adsense-deploy-bak` from the 09-09 TK-11341
+> AdSense fleet deploy (56 G of that a single `wallpapersback.com` full-tree backup).
+>
+> Tracking ticket for the remaining work: **TK-11364**. Evidence:
+> `~/.claude/yolo-queue/verification/TK-10878/outcome-verification-2026-09-10.md`.
+> Everything below this banner is preserved AS WRITTEN for audit — read it through the table above.
+
 **Goal:** bound homesonspec growth (~16 GB/wk) so it never silently returns to an 80 GB crisis, while keeping the full 122 M-row provenance (it's real, unique, live-read — not prunable). This is the "accept the data, make it manageable" path, NOT the 25 GB pruning fantasy.
 
 **Every prod step below is `PROD_DELETE_REQUIRED` / infra-spend → hard-gated. Drafts here; Steve executes on the box.**

← 32416391 Require configured slots for manual AdSense units · back to Homesonspec · ops: make Phase-2 preflight actually enforce the disk gate ( 14ed1904 →