Inspection backlog in Postgres; BullMQ gets a working set

5.4M waiting inspection jobs took 3.3 GB of Redis and most of its single
thread. Producers now add file ids to inspection_queue; index-scheduler
feeds up to INSPECTION_FEED_MAX (50k) into BullMQ. inspection:from-redis
moves the existing backlog over once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commands/IndexScheduler.ts

@@ -5,6 +5,7 @@ import NntpService from '#services/NntpService'
 import QueueService from '#services/QueueService'
 import indexerConfig from '#config/indexer'
 import Shutdown from '#services/Shutdown'
+import InspectionQueue from '#services/InspectionQueue'
 import FetchRangeService from '#services/FetchRangeService'
 import SplitService from '#services/SplitService'
 import BackfillService from '#services/BackfillService'
@@ -108,6 +109,12 @@ export default class IndexScheduler extends BaseCommand {
       }
     }
 
+    // The inspection backlog lives in Postgres; BullMQ gets a working set (LOG.md 2026-10-07 (8)).
+    const feedInspection = async () => {
+      const moved = await InspectionQueue.feed()
+      if (moved > 0) this.logger.info(`Inspection: ${moved} files moved from the backlog to the queue.`)
+    }
+
     const backfill = async () => {
       try {
         const queued = await BackfillService.tick()
@@ -122,7 +129,7 @@ export default class IndexScheduler extends BaseCommand {
     // here once ended the loop while the process stayed up, and live indexing stopped for ~3 hours
     // (LOG.md 2026-10-07 (3)).
     while (!Shutdown.requested) {
-      for (const step of [schedule, backfill, split]) {
+      for (const step of [schedule, backfill, split, feedInspection]) {
         try {
           await step()
         } catch (err: any) {