Batch 1 of Phase 2: migrate 3-4 edge functions to use createEdgeFunction wrapper (process-selective-approval, process-selective-rejection, rate-limit-metrics) to enable automatic error logging, CORS, auth, and reduced boilerplate; preserve existing logic where applicable and prepare for subsequent batches.
Implement Phase 1 by adding error span logging and database persistence to 8 edge functions that already log errors:
- detect-location
- export-user-data
- notify-moderators-submission
- novu-webhook
- send-escalation-notification
- send-password-added-email
- resend-deletion-code
- merge-contact-tickets
This update introduces startSpan/endSpan and logSpanToDatabase usage in catch blocks to ensure errors are recorded in the monitoring database.
Add retry mechanism to backfill-ride-data, backfill-company-data, and backfill-park-locations edge functions using withEdgeRetry, including isRetryableError and isDeadlockError. Apply CORS headers to ride-data and company-data responses, and ensure park-locations already contains CORS remains with retry wrapping. Update unauthorized and success responses to include CORS headers.
Update logger.ts to use the import-map resolved package
- Replace jsr:@supabase/supabase-js@2 with @supabase/supabase-js
- Align edge function logger with deno.json import map for reliable bundling
Add proper CORS headers to edge function:
- Import and apply corsHeadersWithMethods via corsHeaders config
- Include CORS headers in all responses (including unauthorized and success cases)
This resolves preflight and subsequent requests blocked by CORS policy.
Add data quality metrics, growth trends visualization, entity comparison views, and automated health checks to the AdminDatabaseStats dashboard, including new TS types, hooks, UI components, and integrated tabbed layout.
Add admin-only database statistics dashboard
- Introduces types for database statistics and recent additions
- Implements hooks to fetch statistics and recent additions via RPCs
- Adds UI components for stats cards and a recent additions table
- Integrates new AdminDatabaseStats page and routing under /admin/database-stats
- Updates admin sidebar and app routes to expose the new dashboard
- Enables real-time updates and export capabilities for recent additions
Adds comprehensive data completeness dashboard UI and hooks:
- Introduces data completeness types and hook (useDataCompleteness) to fetch and subscribe to updates
- Builds dashboard components (summary, filters, table) and integrates into Admin Settings
- Wireframes for real-time updates and filtering across parks, rides, companies, and ride models
- Integrates into AdminSettings with a new Data Quality tab and route
- Adds data types and scaffolding for analytics, including completeness analysis structure
Adds and fixes a comprehensive data completeness analysis flow:
- Preps migration to create analyze_data_completeness function with weighted scoring
- Addresses security warning by constraining search_path to public schema
- Lays groundwork for real-time updates and integration with admin UI (backfills, filters)
Implement edge function and UI to backfill missing company data from submissions, including admin trigger and result reporting, and wire into admin settings.
Add edge function and UI for ride data backfill, integrating admin UI to trigger backfill, plus supporting types and wiring to backfill_ride_data function. Includes RideDataBackfill component, admin page integration, and initial server-side function scaffold for updating rides from submissions.
Apply migrations and backfill scripts for location data and approval flow, including:
- Implemented backfill_park_locations function to populate missing park locations from approved submissions
- Adjusted process_approval_transaction to properly handle location creation and linking
- Added admin-triggered backfill trigger point and related migration scaffolding for execution
- Created edge-function and admin UI hook to run backfill as needed
- Ensure search_path and relational data usage fixes for stability
Add a script to backfill missing location data for already-approved parks from their submission data, and make it executable from the admin settings interface. Implement a backfill function (SQL migration) and an edge function to trigger it, plus UI integration in Admin Settings to initiate the process. Ensure only parks with approved submissions and location data are linked to new location records, and expose results in admin.
Fix: eliminate approved_at usage in process_approval_transaction and update updated_at instead; ensure updates for submission_items set updated_at and status, remove JSONB-related logic, and recreate function accordingly.
Refactor process_approval_transaction to remove JSONB usage
- Delete JSONB-based building and resolution logic
- Replace with pure relational INSERT/UPDATE paths for parks, rides, companies, ride_models, and photos
- Implement direct temporary reference resolution via SQL queries
- Copy related data (technical specs, coaster stats) after main entity changes
- Remove calls to non-existent functions and simplify flow
Implement Phase 3 audit logging:
- Blog management: create, update, delete, publish
- Manual content deletions (force deletions)
- Direct entity approvals bypassing moderation queue (loggable events where applicable)
Includes integration with central logAdminAction helper and updates to relevant components/hooks:
AdminBlog.tsx (create/update/delete/publish paths)
Moderation queue deletion path (force delete)
Moderation actions paths with direct approvals where possible
New logs for blog_post_created, blog_post_updated, blog_post_deleted, blog_post_published, submission_force_deleted, etc.
Add audit logging for admin settings changes, rate limit config updates, anomaly detection config changes (skipped due to no UI), and version cleanup settings updates. Implement logging via central logAdminAction helper and integrate into AdminSettings, VersionCleanupSettings, and RateLimitAlerts mutations (create, update, delete).
Update PipelineHealthAlerts resolve flow to invalidate all severity-specific system_alerts queries (not just the base key), ensuring resolved alerts disappear from UI. Adds broad invalidation pattern and preserves loading state handling.
Ensure Resolve button clears loading state with proper try-catch-finally around async resolution, preventing stuck loading indicator and improving error handling.
- Introduce per-button loading state in PipelineHealthAlerts
- Disable button and show spinner with "Resolving..." while resolving
- Re-enable after operation completes
- Keeps user feedback inline with alert resolution flow
Enhance PipelineHealthAlerts flow to log and refresh
- Add console logs when resolving alerts on PipelineHealthAlerts
- Invalidate related queries to refresh UI after resolution
- Improve error handling and user feedback paths for resolve action
Investigate and address why the Resolve action isn’t triggering despite superuser role. Add diagnostic logging and ensure the resolve mutation is invoked with correct data, plus prep for validating frontend event wiring and mutation calls.
Implement monitoring and alert-resolve UI improvements:
- Enhance useAlertGroupActions with robust error handling and breadcrumb logging
- Add loading state visuals to GroupedAlertsPanel and Resolve All button
- Integrate loading indicator (Loader2) for better user feedback during resolves
Implement fire-and-forget span persistence:
- Add logSpanToDatabase and persistSpanToDatabase to logger
- Persist spans, attributes, events, and event attributes to new tables
- Wire edgeFunctionWrapper to call DB persistence after each span
- Create required tables, views, and security policies
- Ensure non-blocking and guard for missing Supabase creds
- Change security settings for views to SECURITY INVOKER
- Add explicit search_path in cleanup_old_spans function
- Ensure safe, non-deferring access to trace views and cleanup routine
Update edge function wrapper to emit backend metadata headers (X-Request-Id, X-Span-Id, X-Trace-Id, X-Duration-Ms) on responses; adjust logging to include duration and headers. Enhance edgeFunctionTracking to extract and propagate backend metadata from responses and errors; extend errorHandler to capture and log backend metadata for improved observability.
Migrate the remaining administrative function check-transaction-status to use the wrapEdgeFunction wrapper, aligning with admin-delete-user. Updated to remove manual Deno.serve handling, integrate edge wrapper, standardized logging with edgeLogger, and utilize context for auth and tracing. admin-delete-user already migrated.
Refactor Phase 3 Batch 2–4 Novu-related functions to use the createEdgeFunction wrapper, replacing explicit HTTP servers with edge wrapper, adding standardized logging, tracing, and error handling across subscriber management, topic/notification, and migration/sync functions.
Migrate 6 background jobs to use wrapEdgeFunction: cleanup-old-versions, process-scheduled-deletions, data-retention-cleanup, run-cleanup-jobs, scheduled-maintenance, process-expired-bans. Replace old server routines with edgeFunction wrapper, add centralized logging, tracing, and standardized error handling, and adjust for batch-wise deployment.
Enhance error handling and logging across all edge functions:
- Introduce a shared edgeFunctionWrapper with standardized error handling, request/response logging, tracing, and validation hooks.
- Add runtime type validation utilities (ValidationError, validators, and parse/validate helpers) and integrate into edge flow.
- Implement robust validation for incoming requests and known type mismatches, with detailed logs and structured responses.
- Add post-RPC and post-database error logging to surface type/mismatch issues early.
- Update approval/rejection entry points to leverage new validators and centralized error handling.
Applied migration to fix process_approval_transaction to support all company types: company, manufacturer, operator, property_owner, designer. This resolves unknown item type errors and enables moderation approvals across all entity types. Migration was executed successfully.
Read idempotency key from headers in process-selective-approval and process-selective-rejection, remove it from request bodies, add header validation, and redeploy edge functions to fix idempotency flow.
Improve CORS handling and error logging to fix moderation edge cases:
- Add x-idempotency-key to allowed CORS headers and expose explicit POST methods
- Extend CORS headers to include Access-Control-Allow-Methods
- Update edge function tracing and client error handling to better detect and log CORS/network issues
- Enhance error handling utilities to surface CORS-related failures and provide clearer user messages
Enhance detect-anomalies with advanced ML algorithms (Isolation Forest, seasonal decomposition, predictive modeling) and schedule frequent runs via pg_cron. Updates include implementing new detectors, ensemble logic, and plumbing to run and expose results through the anomaly detection UI and data hooks.
Implements edge function, Django tasks, and UI hooks/panels for automatic retention of old metrics, anomalies, alerts, and incidents, plus updates to query keys and monitoring dashboard to reflect data-retention workflows.
Add Django Celery tasks and utilities to periodically collect system metrics (error rates, response times, queue sizes) and record them into metric_time_series. Include monitoring app scaffolding, metrics collector, Celery beat schedule, middleware for live metrics, and a Supabase edge function for cross-source metrics.
Introduce statistical anomaly detection for metrics via edge function, hooks, and UI components. Adds detection algorithms (z-score, moving average, rate of change), anomaly storage, auto-alerts, and dashboard rendering of detected anomalies with run-once trigger and scheduling guidance.
- Add hooks and components for correlated alerts and incidents
- Integrate panels into MonitoringOverview
- Extend query keys for correlation and incidents
- Implement incident actions (create, acknowledge, resolve) and wiring
