Inactive challenge cleanup

Nightly live-DynamoDB cron that revokes open and direct challenges issued by players inactive ≥ 14 days, pauses matching REALSTANDING presets, and notifies acceptors.

Schedule
Handler src/functions/inactive-challenge-cleanup.ts
Schedule Daily 03:00 UTC (prod only)
Timeout / memory 300 s / 512 MB
Layer gameslib (metaGame UID list for METAGAMES# scan)

Discovery (challenge-centric)

  1. Query pk=USERS (sk, lastSeen) and build inactiveSet (lastSeen < now - 14d; missing lastSeen → active).
  2. Query pk=CHALLENGE (all pending direct challenges).
  3. BatchGet METAGAMES#{metaGame} counts; query STANDINGCHALLENGE#{metaGame} only where standingchallenges > 0.
  4. Filter challenges where challenger.id ∈ inactiveSet.

No S3 dump scan; cost scales with open challenge count, not inactive user count.

Actions per candidate

  1. Re-check issuer USERS.lastSeen (skip if logged in since discovery).
  2. Revoke challenge (mirror node-backend removeAChallenge revocation path).
  3. If standing open challenge matches a REALSTANDING preset entry, set suspended: true on that entry.
  4. Notify acceptors only (email if prefs, push, in-app for direct challenges).

Environment

Variable Default Purpose
ABSTRACT_PLAY_TABLE abstract-play-{stage} DynamoDB table
INACTIVE_CHALLENGE_MS 1209600000 (14d) Inactivity threshold
INACTIVE_CHALLENGE_REVOKE_BATCH_SIZE unlimited Max revokes per run (initial rollout)
VAPID_PUBLIC_KEY / VAPID_PRIVATE_KEY — Web push (optional; skipped if unset)

Preview (read-only, local)

Before the first prod run, preview candidates without writes:

npm run preview-inactive-challenges -- --stage prod

Uses AWS profile AbstractPlayProd / table abstract-play-prod. Optional --days 14 (default).

Manual invoke

serverless invoke -f inactive-challenge-cleanup --stage prod

Use prod with care — this revokes live challenges.

Related