agrs-sequelize-sdk 1.4.42 → 1.4.44

package/models/CampaignCreationLog.js

@@ -292,6 +292,25 @@ module.exports = (sequelize, DataTypes) => {
         {
           fields: ["createdAt"],
         },
+        // Partial index mirroring the V2 side (idx_v2_awaiting_by_feed_campaign).
+        // The hourly Tarzo reconciler runs the SAME awaiting-materialization
+        // WHERE clause against both V1 and V2 (dual-write parity). Without this
+        // index the V1 UPDATE seq-scans 15.8M rows / 11GB and hits the 60s
+        // statement_timeout (verified in prod 2026-07-19: 207s EXPLAIN).
+        //
+        // Composite (feed_campaign_id, createdAt DESC) covers both:
+        //   - step 2/5 join key (rfc.AGRS_CID = l.feed_campaign_id)
+        //   - step 3/5 stale-age filter (l.createdAt < now() - 6h)
+        //
+        // Equality predicate — LIKE would be ignored by the planner (proven
+        // 2026-07-19). Self-pruning: rows drop out on COMPLETED/FAILED flip.
+        {
+          name: "idx_v1_awaiting_by_feed_campaign",
+          fields: ["feed_campaign_id", { name: "createdAt", order: "DESC" }],
+          where: sequelize.literal(
+            `status = 'IN_PROGRESS' AND (details->>'ai_stage') = 'tarzo_awaiting_materialization'`
+          ),
+        },
       ],
     }
   );

package/models/CampaignCreationLogV2.js

@@ -285,28 +285,56 @@ module.exports = (sequelize, DataTypes) => {
         { fields: ["vertical", "feed_provider"] },
         { fields: ["details"], using: "gin" },
         { fields: ["platform_code"] },
-        // Expression index on details->>'ai_stage' for Tarzo-only rows.
-        // Powers the hourly reconciler steps 2/5 (awaiting → COMPLETED) and
-        // 3/5 (stale → FAILED) added by tarzoMaterializationReconciler.
-        // Without this the queries do a full-table scan of the 50GB+ table
-        // and time out (verified in prod 2026-07-19: errors=3 every hour).
-        // Partial WHERE keeps the index tiny — only Tarzo awaiting/failed
-        // rows are indexed, not the whole table.
+        // Partial indexes for the hourly Tarzo reconciler (steps 2/5 and 3/5
+        // in reconcilerService.js). Both steps filter on:
+        //   status='IN_PROGRESS'
+        //   AND details->>'ai_stage'='tarzo_awaiting_materialization'
+        // Predicate is equality — NOT `LIKE 'tarzo_%'` — because Postgres's
+        // partial-index prover doesn't cross equality→LIKE, so a LIKE-predicate
+        // index would be ignored (verified in prod 2026-07-19). Equality
+        // matches the queries verbatim, so the planner picks these up.
+        //
+        // Column choices:
+        //   - feed_campaign_id: drives the nested-loop join to
+        //     rsoc_feed_campaigns.AGRS_CID in step 2/5.
+        //   - created_at: powers the stale age filter in step 3/5.
+        //
+        // Self-pruning: rows leave the index as soon as status flips to
+        // COMPLETED/FAILED, so index size stays proportional to in-flight
+        // rows only.
         {
-          name: "idx_v2_ai_stage_tarzo",
-          fields: [sequelize.literal(`((details->>'ai_stage'))`)],
-          where: sequelize.literal(`(details->>'ai_stage') LIKE 'tarzo_%'`),
+          name: "idx_v2_awaiting_by_feed_campaign",
+          fields: ["feed_campaign_id"],
+          where: sequelize.literal(
+            `status = 'IN_PROGRESS' AND (details->>'ai_stage') = 'tarzo_awaiting_materialization'`
+          ),
         },
-        // Composite index for the stale-cleanup query (step 3/5) which
-        // filters by ai_stage AND created_at < now() - 6h. Same partial
-        // WHERE keeps size minimal.
         {
-          name: "idx_v2_ai_stage_tarzo_created_at",
-          fields: [
-            sequelize.literal(`((details->>'ai_stage'))`),
-            { name: "created_at", order: "DESC" },
-          ],
-          where: sequelize.literal(`(details->>'ai_stage') LIKE 'tarzo_%'`),
+          name: "idx_v2_awaiting_created_at",
+          fields: [{ name: "created_at", order: "DESC" }],
+          where: sequelize.literal(
+            `status = 'IN_PROGRESS' AND (details->>'ai_stage') = 'tarzo_awaiting_materialization'`
+          ),
+        },
+        // Partial index for the ai_campaign_queue backfill query in step 2/5
+        // (completeAdoptedAwaitingLogs's third UPDATE). After the awaiting log
+        // flips to COMPLETED, ai_stage becomes 'tarzo_materialized_and_adopted'
+        // and the queue's facebook_campaign_id gets backfilled by joining V2
+        // to ai_campaign_queue via process_id.
+        //
+        // Without this, the query bitmap-scans by status='COMPLETED' (returns
+        // 1.2M rows), then filters ai_stage in memory (rejects 1.18M) → 175s
+        // seq scan, 60s timeout in prod (verified 2026-07-19).
+        //
+        // Composite (process_id, feed_campaign_id) supports both the join to
+        // ai_campaign_queue.creation_logs_process_id AND the further join to
+        // rsoc_feed_campaigns.AGRS_CID.
+        {
+          name: "idx_v2_adopted_by_process",
+          fields: ["process_id", "feed_campaign_id"],
+          where: sequelize.literal(
+            `status = 'COMPLETED' AND (details->>'ai_stage') = 'tarzo_materialized_and_adopted'`
+          ),
         },
       ],
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agrs-sequelize-sdk",
-  "version": "1.4.42",
+  "version": "1.4.44",
   "main": "index.js",
   "scripts": {
     "start": "node index.js",