Fix daily content cron: IG mirror silently failing since 2026-05-27

[neverpeakev]

Diagnosis

Routine health check on 2026-07-08 found today's FB post published cleanly (gen-2026-07-08-p9hrc, fb_post_id 1086292991223812_122116418852989743) but the IG mirror row still shows status='fb_published' with ig_media_id=null and error_message=null. The IG cron ran at 14:15:08 UTC and returned 200 with no log output.

Pulling recent history from generated_posts:

• Last successful IG publish: 2026-05-27 (42 days ago).
• Every daily post from 2026-06-29 → 2026-07-08 is stuck at fb_published with ig_published_at=null, error_message=null. Ten rows total sitting idle.
• Immediately preceding failures (2026-05-23/24/26) were correctly recorded as ig_failed with Meta's code=100, error_subcode=33, "Authorization Error" — the token needed re-authorization or a missing scope.

What happened: after 2026-05-27, the failure point moved from inside the per-row publishToIg call (which correctly caught + marked ig_failed) to inside fetchPageContext, which runs once before the row loop. That catch block returned 200 with ok:false in the body but wrote no console.error and did not touch any row — so 42 days of failures were invisible to both Vercel runtime logs and the DB. This routine check only caught it because it cross-referenced ig_published_at history.

Almost certainly the same underlying Meta scope/token regression as the late-May ig_failed rows — the page-token likely lost instagram_basic and/or instagram_content_publish when the system-user token was last rotated — but the observability gap prevented anyone from noticing.

Changes

app/api/social/ig-mirror/route.ts

• console.error at both failure branches (auth/page-context and the per-row publish catch) so future Meta errors surface in Vercel runtime logs.
• When fetchPageContext throws, mark each row in pending as ig_failed with the error message. Previously the rows were left untouched and re-queued indefinitely, so the failure was doubly hidden (no log, no DB trace). Now the DB carries the diagnosis and the next routine run can grep error_message.

Immediate follow-up (operator)

After merge, the next 14:15 UTC cron will mark today's stuck row (and any others) as ig_failed with the actual Meta response in error_message. Read that error, then either:

1. Re-issue META_MARKETING_API_TOKEN with instagram_basic + instagram_content_publish scopes, or
2. If the IG business account got unlinked from the FB page, re-link it in Meta Business Suite.

Once the token is fixed, the 10 stuck rows can be backfilled by clearing their status back to fb_published (they're already ig_failed after this PR runs once, so the listPendingIgMirror filter would skip them).

Test plan

• Merge, wait for next 14:15 UTC cron, confirm Vercel runtime logs show [ig-mirror] auth/page-context failed: with the Meta error body.
• Confirm today's row transitions from fb_published → ig_failed with a populated error_message.
• Fix Meta scopes, then verify next day the row lands as ig_published with an ig_media_id.

---

Generated by Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
momfluence	Ready	Preview, Comment	Jul 8, 2026 2:53pm
momfluence-platform	Ready	Preview, Comment	Jul 8, 2026 2:53pm