bwi writer migration whey→venus (PG + wi-http authoritative on venus; whey → backup-only)

Sub-tickets — 7/7 done

Questions

Activity

• wi-cli-whey created · 7w ago
• wi-cli-whey commented · 7w ago
  Elazar-approved (infra-coders audit, 2026-06-26). venus PG17 up, greenfield db/role create. Ingress: same bwi.pensanta.com — Caddy+DNS flip on venus owned by bin-venus on GO. whey current: wi-http active 127.0.0.1:8042→local PG; bwi-venus-tunnel inactive. Part of venusinfra (VENUSINF) program. Children: BWI-12 (venus db/role) BWI-13 (wi-http deploy) BWI-14 (dump/restore parity) BWI-15 (cutover) BWI-16 (client repoint+Caddy/DNS) BWI-17 (reverse-repl+quiesce whey) BWI-18 (validate). Roles: bin-venus=host/net/Caddy; db-basquetwi=PG/dump; coder-basquetwi=wi-http/env/repoint; pm=sequencing+signoff.
• wi-cli-whey commented · 7w ago
  INCIDENT: migration premise wrong. NOT whey-authoritative+greenfield-venus. Verified: venus wi-http live+public since 06-19 (bwi.pensanta.com=91.99.136.171), own PG ~1336 WI, whey loopback PG ~1160, both live-diverging ~6d. Agent CLI(whey)→whey PG; public/off-host→venus PG. Whole plan reframes to RECONCILE two diverged DBs, not copy. All stop/drop/restore frozen. coder mapping divergence; Elazar escalated for reconcile decision.
• wi-cli-whey commented · 7w ago
  Refined: whey=CANONICAL for active backlog (has BWI-11..18 + all agent writes via ZT base→whey PG). venus=divergent lineage: BEHIND on BWI (stops #10) but AHEAD +176 WI in OTHER projects + live public-UI write trickle (latest 02:24 today). Asymmetric divergence from a common ancestor. Path hinges on id-collision: if id spaces don't collide → merge venus-uniques INTO whey (superset) then whey→venus replica = zero loss; if collide → renumber/discard. coder producing id-space map + venus-only export. Elazar escalated. All frozen.
• wi-cli-whey commented · 7w ago
  Divergence map (coder, verified by PM via /tmp/bwi-venus-divergence.tsv): fork=id1151; ids1..1151 identical on (id,createdAt,title); whey post-fork uniques=9 (BWI-11..18+monitor-rename); venus post-fork uniques=185 REAL cross-project items via public-UI (PLUTO57/MARS45/MSG33/EVO31/VENUS13/OPS3/AYUD3), through 02:20 today. KEY-N refs: 184 free-merge, 1 clash (OPS-21). MERGE feasible, near-lossless. Direction chosen: VENUS-SUPERSET (merge whey's 9 into venus = the 24/7 authoritative target; fewer moves) — coder validating. OPEN risk: mutable-field/event edits on the 1151 shared items not yet quantified (coder). Then single-endpoint convergence: venus authoritative, whey clients repoint, whey→backup. Elazar updated. Still frozen.
• wi-cli-whey commented · 7w ago
  FINAL PLAN (pending Elazar GO): shared-base reconcile=6 items, venus ahead on ALL 6, whey never ahead → venus is the more-current copy. whey uniques=9 only. Approach=VENUS-SUPERSET: graft whey's 9 (+events) into venus, renumber OPS-21, bump venus BWI nextWorkItemNumber 11→19. Then repoint clients to https bwi.pensanta.com/api (fleet-wide incl lezama), quiesce whey→backup. Seq: freeze whey writes→graft→repoint→demote. coder writing transform, db reviews, execution frozen until Elazar GO. BWI-14 canceled (superseded). Elazar sent GO-request.
• wi-cli-whey statusChanged · 7w ago
  Elazar GO'd all 3 (venus-authoritative / whey write-freeze / public-https repoint). Executing cutover: prep (transform review + venus key-parity) → freeze whey writes → graft 9 → repoint clients to https → demote whey to backup.
• wi-cli-whey completed · 7w ago
  MIGRATION COMPLETE. venus authoritative (https://bwi.pensanta.com/api), whey backup-only. All children terminal: BWI-12/13/15/16/17/18 done, BWI-14 canceled. Validated e2e + reverse backup verified.