Reliability, Scaling and Troubleshooting Volume 07 - Advanced Application Architecture
Realtime is lossy from the app perspective Clients can sleep, lose Wi-Fi, change networks, suspend background tabs, or disconnect during token expiry. Therefore TaskFlow never assumes it observed every event. Reconnect means: rejoin + refetch authoritative state.
Subscription statuses • SUBSCRIBED - channel is joined; refresh current data.
• CHANNEL_ERROR - inspect the full error payload.
• TIMED_OUT - connectivity/join problem.
• CLOSED - channel ended; determine whether expected cleanup or unexpected disconnect.
Built-in reconnection The Realtime protocol/client supports reconnection with backoff. Application code should still make recovery idempotent and refresh server state after re-subscription.
Scaling lesson Supabase currently recommends Broadcast for database-change subscriptions because Postgres Changes can become costly under many subscribers; each database change can require authorization work across subscribers. Measure your own workload and plan limits.
Troubleshooting order • Confirm the user has a valid Supabase session/JWT.
• Confirm setAuth() occurs before joining a private channel.
• Confirm topic string exactly matches the policy.
• Confirm Membership/authSubject mapping.
• Confirm Realtime public-access/private-channel setting.
• Confirm trigger exists and fires.
• Check Realtime logs and browser WebSocket frames.
• Confirm removeChannel cleanup to avoid duplicate subscriptions.
Observability Log channel topic (not token), status, Workspace id, reconnect count, error category and request/user correlation where policy allows. Never log access tokens.
Reliability Scaling and Troubleshooting matters because it changes a real boundary in TaskFlow. The goal is to understand the inputs, owner, observable output, and failure modes before adding more abstraction.
Restate the goal of “Reliability Scaling and Troubleshooting” as one observable behavior. Write the success signal before touching code.
Trace the boundary map: database change → event stream → subscription → cache reconciliation → UI. Mark the one boundary this lesson changes.
Inspect the current repository state and record the exact files, route, table, or runtime involved. Avoid creating an abstraction before you can name the existing owner.
Implement the smallest vertical slice. Add one coherent change, run the narrowest verification command, and read the output before continuing.
Connect the slice to TaskFlow’s phase artifact: Realtime task updates that merge into TanStack Query state with scoped subscriptions and reconnect behavior.. Verify behavior from the outside as a user or API client would.
Create one controlled failure that is safe to reproduce. Predict the symptom first, capture evidence, locate the earliest broken boundary, then fix only that cause.
Close the reference and rebuild or explain the core change from memory. Record what you would monitor or test before trusting the same change in production.
channel.on("postgres_changes", filter, payload => {
queryClient.invalidateQueries({ queryKey: taskKeys.all });
});Do not copy this as a finished answer. Predict what each boundary does, type the smallest useful slice, run it, then compare the result with your prediction.
Return to the nearest observable boundary: input, network, process log, database, or rendered UI. Compare actual evidence with the predicted path.
Change one variable at a time and reproduce the original failure after the fix; otherwise you have correlation, not diagnosis.
Without copying the final source, create a second small scenario that applies “Reliability Scaling and Troubleshooting” to a different TaskFlow resource or input. Write the acceptance criteria first, implement it, and provide the evidence that each criterion is satisfied.
Test disconnect/reconnect and ensure a fresh server query repairs any missed events.
Use the smallest change that demonstrates this lesson. Stop after it works and capture evidence.Events for other tenants visible to a client
Predict the symptom before causing the failure. Then locate the earliest boundary where reality diverges from expectation.Why does this responsibility live here, what input does it trust, what output proves success, and what would go wrong if another layer owned it?
If your explanation is vague, repeat the boundary trace before moving on.Close the reference. Reproduce the core idea from memory or explain the exact request/data flow aloud. Then check your diff or notes for accidental shortcuts that bypass validation, authorization, error handling, or verification earned in earlier phases.