feat(docs): add resumable WebSockets + Edge Functions troubleshooting guides (#46178)

## I have read the
[CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md)
file.

YES

## What kind of change does this PR introduce?

Docs update (new guides + follow-up documentation fix from review
feedback).

## What is the current behavior?

There was no consolidated docs example for resumable WebSockets with
Edge Functions, and no dedicated troubleshooting guide for worker
timeouts / WebSocket drops.

## What is the new behavior?

- Adds a resumable WebSockets guide for Edge Functions, including:
  - session persistence
  - event replay
  - idempotency pattern and schema examples
  - client/server example flow
- Adds an Edge Functions troubleshooting guide for worker timeouts and
WebSocket drops.
- Updates docs navigation to surface the new guides.
- Follow-up fix from review feedback: the browser client example now
stores `sessionId` and `lastEventId` in `sessionStorage` (instead of
`localStorage`).

## Additional context

- Branch has been updated with latest `origin/master`.
- This PR remains documentation-focused; no production runtime code
changes were introduced.

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Documentation**
* Added a guide on resumable WebSockets covering session persistence,
event replay, idempotency patterns, SQL schema examples, and
client/server usage.
* Added a troubleshooting guide on Edge Functions worker timeouts and
WebSocket drops with scenarios, symptoms, and practical workarounds.
* Enhanced WebSocket docs with a production note on worker lifecycle and
keeping runtime promises open to avoid premature shutdown.
  * Navigation updated to surface the new guides.

<!-- review_stack_entry_start -->

[![Review Change
Stack](https://storage.googleapis.com/coderabbit_public_assets/review-stack-in-coderabbit-ui.svg)](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46178?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack)

<!-- review_stack_entry_end -->

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Co-authored-by: Lakshan Perera <lakshan@supabase.io>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
Co-authored-by: CodeRabbit <noreply@coderabbit.ai>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
This commit is contained in:
authored and GitHub committed 2026-05-27 18:34:23 +00:00
1 parent 42f1f19fdd
commit 894cb531d1
4 files changed
+406

No files matched your search

@@ -1656,6 +1656,10 @@ export const functions: NavMenuConstant = {
name: 'Troubleshooting',
url: '/guides/functions/troubleshooting' as `/${string}`,
},
{
name: 'Worker timeouts and WebSocket drops',
url: '/troubleshooting/edge-functions-worker-timeouts-and-websocket-drops' as `/${string}`,
},
],
},
{
@@ -1783,6 +1787,10 @@ export const functions: NavMenuConstant = {
name: 'Image Transformation & Optimization',
url: '/guides/functions/examples/image-manipulation' as `/${string}`,
},
{
name: 'Resumable WebSockets with replay',
url: '/guides/functions/examples/resumable-websockets' as `/${string}`,
},
],
},
{
@@ -0,0 +1,225 @@
---
title: 'Resumable WebSockets with Edge Functions'
description: 'Build reconnect-safe WebSockets with event replay, idempotency keys, and graceful restarts.'
---
This example shows how to build a reconnect-safe chat stream on Supabase Edge Functions using:
- WebSocket upgrade + JWT auth
- Postgres-backed session and event persistence
- Event replay with `lastEventId`
- Idempotent user messages with `idempotency_key`
- Graceful client reconnects during worker restarts
Reference implementation: [Building Resumable WebSockets with Supabase Edge Functions and Postgres](https://blog.mansueli.com/building-resumable-websockets-with-supabase-edge-functions-and-postgres)
## Architecture
1. Client connects with a user JWT, plus optional `sessionId` and `lastEventId`.
2. Function verifies auth and either resumes or creates a session.
3. Every message is written to `ws_events` with an incrementing `id`.
4. On reconnect, server replays events where `id > lastEventId`.
5. Client updates local `lastEventId` and resumes without losing messages.
## Database schema
```sql
create extension if not exists pgcrypto;
create table ws_sessions (
id uuid primary key default gen_random_uuid(),
user_id uuid not null,
created_at timestamptz default now(),
updated_at timestamptz default now(),
last_event_id bigint default 0
);
create table ws_events (
id bigint generated by default as identity primary key,
session_id uuid not null references ws_sessions(id) on delete cascade,
event_type text not null,
payload jsonb not null,
created_at timestamptz default now()
);
create index ws_events_session_id_id_idx on ws_events(session_id, id);
create table ws_idempotency_keys (
session_id uuid not null references ws_sessions(id) on delete cascade,
idempotency_key uuid not null,
primary key(session_id, idempotency_key)
);
create unlogged table ws_live_connections (
session_id uuid primary key,
connected_at timestamptz default now(),
last_seen_at timestamptz default now(),
edge_region text
);
```
## Edge Function (WebSocket proxy)
Use `supabase functions serve --no-verify-jwt` and validate JWT inside the function.
```ts
import { createAdminClient, createContextClient, verifyCredentials } from '@supabase/server/core'
const PREEMPTIVE_RESTART_MS = 340_000
function send(socket: WebSocket, payload: unknown) {
if (socket.readyState === WebSocket.OPEN) {
socket.send(JSON.stringify(payload))
}
}
Deno.serve(async (req) => {
const url = new URL(req.url)
const token = url.searchParams.get('token')
if (!token) return new Response('Missing token', { status: 401 })
const { data: auth, error } = await verifyCredentials({ token, apikey: null }, { auth: 'user' })
if (error || !auth?.userClaims?.id) {
return new Response('Unauthorized', { status: 401 })
}
const admin = createAdminClient()
const { socket, response } = Deno.upgradeWebSocket(req, { idleTimeout: 0 })
// Prevent EarlyDrop by keeping a pending promise until socket close.
let resolveClosed!: () => void
const closed = new Promise<void>((resolve) => {
resolveClosed = resolve
})
// @ts-ignore
EdgeRuntime.waitUntil(closed)
const requestedSessionId = url.searchParams.get('sessionId')
const lastEventId = Number(url.searchParams.get('lastEventId') || 0)
const sessionId = requestedSessionId ?? crypto.randomUUID()
socket.onclose = () => {
resolveClosed()
}
socket.onmessage = async (event) => {
const msg = JSON.parse(event.data)
if (msg.type === 'user_message') {
const { error: idempotencyError } = await admin.from('ws_idempotency_keys').upsert(
{
session_id: sessionId,
idempotency_key: msg.idempotency_key,
},
{ onConflict: 'session_id,idempotency_key', ignoreDuplicates: true }
)
let userEvent
if (idempotencyError) {
// Conflict detected - this is a retry, fetch the existing event
const { data: existingEvent } = await admin
.from('ws_events')
.select()
.eq('session_id', sessionId)
.eq('idempotency_key', msg.idempotency_key)
.single()
userEvent = existingEvent
} else {
// New idempotency key - insert the event
const { data: newEvent } = await admin
.from('ws_events')
.insert({
session_id: sessionId,
event_type: 'user_message',
payload: { content: msg.content },
})
.select()
.single()
userEvent = newEvent
}
send(socket, {
type: 'user_message',
payload: userEvent?.payload,
event_id: userEvent?.id,
})
}
}
send(socket, { type: 'session_init', session_id: sessionId })
queueMicrotask(async () => {
const { data: replayEvents } = await admin
.from('ws_events')
.select('*')
.eq('session_id', sessionId)
.gt('id', lastEventId)
.order('id')
for (const event of replayEvents ?? []) {
send(socket, {
type: event.event_type,
payload: event.payload,
event_id: event.id,
replay: true,
})
}
})
setTimeout(() => {
send(socket, { type: 'server_restarting' })
socket.close(1012, 'Service restart')
}, PREEMPTIVE_RESTART_MS)
return response
})
```
## Browser client
The client stores `sessionId` and `lastEventId` in session storage, then reconnects with exponential backoff.
```ts
let sessionId = sessionStorage.getItem('ws_session_id')
let lastEventId = Number(sessionStorage.getItem('last_event_id') || 0)
function connect(token: string) {
const url =
`wss://YOUR_PROJECT.functions.supabase.co/websocket-proxy` +
`?token=${encodeURIComponent(token)}` +
`&lastEventId=${lastEventId}` +
(sessionId ? `&sessionId=${sessionId}` : '')
const ws = new WebSocket(url)
ws.onmessage = (e) => {
const msg = JSON.parse(e.data)
if (msg.event_id) {
lastEventId = Math.max(lastEventId, msg.event_id)
sessionStorage.setItem('last_event_id', String(lastEventId))
}
if (msg.type === 'session_init') {
sessionId = msg.session_id
sessionStorage.setItem('ws_session_id', sessionId)
}
}
}
```
## Why this pattern works
- If the worker restarts, the client reconnects with the same session.
- Replay closes delivery gaps caused by reconnect windows.
- Idempotency keys prevent duplicate inserts when clients retry.
- `EdgeRuntime.waitUntil()` prevents unexpected early termination of idle-looking WebSocket workers.
## Next steps
- Add row-level security policies for all `ws_*` tables.
- Add a heartbeat and cleanup policy for stale sessions.
- Add structured event payload types and input validation.
- Add observability dashboards for disconnect rate and replay lag.
@@ -13,6 +13,8 @@ This allows you to:
- Create WebSocket relay servers for external APIs
- Establish both incoming and outgoing WebSocket connections
For a production-ready reconnect pattern with session persistence and replay, see [Resumable WebSockets with Edge Functions](/docs/guides/functions/examples/resumable-websockets).
---
## Creating WebSocket servers
@@ -249,6 +251,8 @@ The maximum duration is capped based on the wall-clock, CPU, and memory limits.
</Admonition>
When using WebSockets, keep in mind that the HTTP request is considered complete after `Deno.upgradeWebSocket(req)` returns the response. To prevent early worker retirement while the socket is still open, keep an unresolved `EdgeRuntime.waitUntil()` promise that resolves in `socket.onclose`.
---
## Testing WebSockets locally
@@ -0,0 +1,169 @@
---
title = "Edge Functions worker timeouts and WebSocket drops"
topics = [ "functions" ]
keywords = [ "websocket", "timeout", "earlydrop", "wall clock", "cpu limit", "streaming", "cold start" ]
teams = [ "team-functions", "team-support" ]
types = [ "support" ]
---
## Background
Edge Functions run inside V8 isolates managed by a supervisor in `edge-runtime`.
The supervisor enforces resource limits:
- Wall clock time (`worker_timeout_ms`)
- CPU time (soft + hard limits)
- Memory usage
It may retire early (`EarlyDrop` event) if the isolate is idle Isolate is considered idle if following conditions are met:
- The HTTP response has already been returned.
- All `EdgeRuntime.waitUntil()` promises have resolved.
If both are true during a resource check, the isolate can be terminated even with open WebSocket connections.
## Scenario 1: WebSocket drops around half of the wall clock limit
### Symptoms
- WebSocket closes at a consistent interval (often around half of the configured wall clock limit).
- Logs include `EarlyDrop` or wall clock warning events.
### Root cause
After `Deno.upgradeWebSocket(req)` returns a response, the HTTP request is considered acknowledged. If there is no unresolved `waitUntil` work, the worker may look idle and be retired early.
### What to check
1. Compare connection lifetime to your wall clock limit.
2. Inspect logs for `EarlyDrop` around disconnect time.
3. Confirm there is no unresolved `EdgeRuntime.waitUntil()` promise tied to socket lifecycle.
### Workaround
Keep a promise pending until the socket closes.
```ts
Deno.serve((req) => {
const { socket, response } = Deno.upgradeWebSocket(req)
const socketClosedPromise = new Promise<void>((resolve) => {
socket.onclose = () => resolve()
})
EdgeRuntime.waitUntil(socketClosedPromise)
socket.onmessage = (event) => {
socket.send(event.data)
}
return response
})
```
`EdgeRuntime.waitUntil()` prevents early retirement, but it does not extend the hard wall clock limit.
## Scenario 2: Function killed at a consistent duration
### Symptoms
- Function fails at a predictable runtime (for example, always around the same second mark).
- Logs include wall clock shutdown reasons.
- Clients may receive status `546` or cancellation errors.
### Root cause
The function exceeded the configured wall clock budget.
### Workarounds
- Split work into smaller units.
- Move long work to async/background processing and return early.
- Use streaming from upstream APIs where possible.
- Use queues (`pg_net`, `pgmq`, or webhooks) for chunked processing.
## Scenario 3: Function killed by CPU limit
### Symptoms
- Failures during compute-heavy tasks.
- Logs include CPU soft/hard limit events.
### Root cause
CPU budget and wall clock budget are independent. A function can run out of CPU time long before wall clock is exhausted.
### Workarounds
- Break large synchronous loops into async chunks.
- Optimize expensive paths and avoid repeated recalculation.
- Move heavy compute to systems designed for long CPU-bound workloads.
## Scenario 4: SSE or AI streams end before completion
### Symptoms
- Streaming starts but ends prematurely.
- No final `[DONE]` token or normal close marker.
### Root cause
The worker hits wall clock or early retirement conditions while forwarding a long stream.
### Workaround
Keep the isolate alive for the stream piping lifecycle:
```ts
Deno.serve(async (_req) => {
const upstream = await fetch('https://api.openai.com/v1/chat/completions', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
Authorization: `Bearer ${Deno.env.get('OPENAI_API_KEY')}`,
},
body: JSON.stringify({ stream: true }),
})
const { readable, writable } = new TransformStream()
EdgeRuntime.waitUntil(upstream.body!.pipeTo(writable))
return new Response(readable, {
headers: { 'Content-Type': 'text/event-stream' },
})
})
```
## Scenario 5: Cold starts fail before first response
### Symptoms
- First request after idle fails (for example, `504` or worker creation timeout).
- Subsequent requests may succeed.
### Root cause
Large dependency trees or expensive top-level initialization can exceed startup budget.
### Workarounds
- Avoid slow top-level `await` work.
- Lazy-initialize heavy clients inside request handlers.
- Reduce bundle and dependency size.
- Keep critical functions warm if needed.
## Key distinctions
- `EarlyDrop`: early retirement when worker appears idle.
- `WallClockTime`: hard runtime ceiling reached.
- CPU and wall clock limits are independent.
- WebSockets are not request-tracked by the supervisor after upgrade.
## Related resources
- [Edge Functions limits](/docs/guides/functions/limits)
- [Edge Function shutdown reasons explained](./edge-function-shutdown-reasons-explained)
- [Monitoring Edge Function resource usage](./edge-function-monitoring-resource-usage)
- [Handling WebSockets in Edge Functions](/docs/guides/functions/websockets)