EnableJob does not compute next_run_at — enabled cron jobs never fire

[duhd-vnpay]

Bug

PGCronStore.EnableJob() sets enabled = true but does not update next_run_at. Since GetDueJobs() filters by next_run_at <= now, the job never fires after being re-enabled via the UI toggle (cron.toggle RPC).

Root Cause

// cron_crud.go — current code
func (s *PGCronStore) EnableJob(jobID string, enabled bool) error {
    _, err = s.db.Exec("UPDATE cron_jobs SET enabled = $1, updated_at = $2 WHERE id = $3", enabled, time.Now(), id)
    // ...
}

When enabling: next_run_at stays NULL (cleared when disabled), so GetDueJobs skips it forever. recomputeStaleJobs() only runs at startup, not after UI toggle.

When disabling: next_run_at is not cleared, so the scheduler may still pick it up briefly.

Expected Behavior

• Enable: read schedule fields from DB, call computeNextRun(), set next_run_at
• Disable: set next_run_at = NULL immediately

Suggested Fix

func (s *PGCronStore) EnableJob(jobID string, enabled bool) error {
    if !enabled {
        // Clear next_run_at so GetDueJobs skips immediately
        s.db.Exec("UPDATE cron_jobs SET enabled = false, next_run_at = NULL, updated_at = $1 WHERE id = $2", now, id)
        s.cacheLoaded = false
        return nil
    }
    // Enable: read schedule, compute next_run_at
    row := s.db.QueryRow(`SELECT schedule_kind, cron_expression, run_at, timezone, interval_ms FROM cron_jobs WHERE id = $1`, id)
    // ... scan into CronSchedule ...
    next := computeNextRun(&schedule, now, defaultTZ)
    s.db.Exec("UPDATE cron_jobs SET enabled = true, next_run_at = $1, updated_at = $2 WHERE id = $3", next, now, id)
    s.cacheLoaded = false
    return nil
}

Steps to Reproduce

1. Create a cron job (e.g., 0 7 * * *)
2. Disable it via UI → cron.toggle RPC
3. Re-enable it via UI → cron.toggle RPC
4. Wait for scheduled time — job never fires
5. Check DB: next_run_at is NULL despite enabled = true

Environment

• GoClaw v1.74.1
• PostgreSQL-backed cron store (PGCronStore)

[mrgoonie]

Clear root cause analysis. The EnableJob() path not recomputing next_run_at means re-enabled jobs are effectively dead.

Your suggested fix looks correct - enable should compute next_run_at from the schedule, and disable should NULL it. The cacheLoaded = false reset is a good touch to force a cache refresh.

Open to a PR on this - it's a focused fix.