Supabase Problems & Issues

Free projects with low activity over 7 days are paused automatically; paused projects can be restored from the dashboard for up to a year. Users report demos breaking and run keep-alive cron jobs to avoid it. Free projects also have no backups.

Users report being charged $300+ when expecting $25/month due to 'ghost compute' resources. Billing for resources users weren't aware they were using. Support often unresponsive while charges accumulate. Small unpaid bills ($16) result in immediate project suspension.

Pro is $25 but each project is billed its own compute; the $10 credit covers only one Micro instance, so a second project adds $10+ and larger instances run $60-$410/month. Egress allowance rose to 250 GB on Pro, but MAU, disk, PITR ($100/month) and add-ons still push small apps past $100/month.

The jump from Pro ($25/month) to Team ($599/month) is huge, and Team keeps the same usage quotas as Pro - the money buys SOC 2, ISO 27001, scoped roles, SSO and longer backups/logs. Uptime SLAs need Enterprise with custom pricing.

No live chat support even on Enterprise plans, which feels outdated in 2026. No phone support at any tier. Free users get Discord community support with variable response times. Complex Postgres issues get bounced back as 'database questions, not Supabase questions.'

Row Level Security is powerful but performance impact is significant, especially for queries scanning every row. Calling functions in RLS policies evaluates for each row, causing slowdowns. Complex workflows still need app-level checks. Learning curve is steep for developers.

The Supabase CLI for local development has broken Docker containers, migration system issues, and breaks frequently. Val Town reported the CLI couldn't be reliably used - migrations failed, containers wouldn't start. Forces developers to test against remote database.

Each compute instance has maximum database connections and connection pooler limits. Real-world reports of queries taking 8+ seconds, database hitting connection limits, and real-time subscriptions dropping constantly as user base grows. Throttling occurs when IOPS limits hit.

Edge Functions have 60-second execution limit and 2-second CPU time per request. Cold starts historically ranged from 400ms to 1.2s. Web Worker API not available. Heavy long-running jobs must be moved to background workers. Not suitable for all serverless use cases.

Status trackers log hundreds of Supabase incidents since 2021. In September 2026 alone: Nano projects became unresponsive after long uptime (fixed by manual restart), a Dashboard/permissions incident and an eu-west-1 hardware failure; in August project creation failed for ~2.7 hours.

Subscription limit exceeded errors when active subscriptions surpass plan limits. Connection limits cause WebSocket drops. JavaScript timer throttling prevents heartbeats, causing silent disconnections. High-concurrency apps hit these limits quickly.

The built-in email sender is only meant for testing and allows a handful of auth emails per hour, so confirmation and password-reset emails silently stop once real users sign up. Custom SMTP is effectively mandatory for production.

500 errors in Auth typically indicate database or SMTP provider issues. Email link authentication commonly fails with 'Database error finding user from email link.' Session timeouts and bugs cause work loss. Auth-helpers package being deprecated adds migration burden.

Documentation is inconsistent and can frustrate developers navigating advanced features. Lack of easy access logs makes debugging database issues challenging. Knowledge base needs more troubleshooting guides for edge cases. Auth documentation especially problematic.

The direct host db.<ref>.supabase.co resolves only to IPv6, so many hosts (Railway, Fly regions, home networks, some CI) fail with 'Network is unreachable'. The free fix is the Supavisor pooler string; a dedicated IPv4 address is a paid add-on.