openclaw - 💡(How to fix) Fix Gateway crashes on provider WebSocket disconnect (code 1005) — no reconnect attempt [2 comments, 1 participants]

openclaw2026-03-29 05:17:27

ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

GitHub issue URL

Helpful · Quick feedback

GitHub stats

openclaw/openclaw#56816•Fetched 2026-04-08 01:47:31

View on GitHub

Comments

Participants

Timeline

Reactions

Author

henrykkng

Participants

henrykkng

Timeline (top)

commented ×2

The gateway process crashes and restarts every ~35 minutes when the provider WebSocket connection (OpenRouter) drops with code 1005 (no status received). The SafeGatewayPlugin has maxReconnectAttempts: 0, so instead of attempting reconnection, it throws an uncaught exception that kills the entire gateway process.

Error Message

Happy to provide full log excerpts or additional debugging if needed.

Root Cause

OpenRouter (and likely other providers) periodically cycles WebSocket connections — this is normal behavior
WebSocket code 1005 means "no status received" — the remote end closed without a proper close frame
SafeGatewayPlugin is configured with maxReconnectAttempts: 0 — any connection drop triggers a crash
The plugin does not distinguish between "provider unavailable" and "transient connection drop"

Code Example

$ grep "Uncaught exception" openclaw-2026-03-29.log | wc -l
18  ← 18 crashes in one day

---

11:24:40 — crash #17
11:59:48 — crash #18  (~35 min gap)

---

[openclaw] Uncaught exception: Error: Max reconnect attempts (0) reached after code 1005
    at SafeGatewayPlugin.handleReconnectionAttempt (file:///.../provider-CAlWEl41.js:3318:47)
    at SafeGatewayPlugin.handleClose (file:///.../provider-CAlWEl41.js:3364:8)
    at WebSocket.<anonymous> (file:///.../provider-CAlWEl41.js:3307:9)

---

$ ip route get 172.67.24.1
172.67.24.1 via 192.168.1.1 dev eno1  ← direct internet, not Tailscale

$ curl -so /dev/null -w "connect: %{time_connect}s, tls: %{time_appconnect}s\n" https://openrouter.ai
connect: 0.007s, tls: 0.038s  ← fast direct connection

RAW_BUFFERClick to expand / collapse

Description

Environment

OpenClaw version: (current as of 2026-03-29)
OS: Linux 6.8.0-106-generic (Ubuntu, x64)
Node: v22.22.1
Provider: OpenRouter (nvidia-qwen122/qwen/qwen3.5-122b-a10b)
Gateway config: loopback bind, systemd service

Evidence

$ grep "Uncaught exception" openclaw-2026-03-29.log | wc -l
18  ← 18 crashes in one day

Crashes occur every ~35 minutes consistently:

11:24:40 — crash #17
11:59:48 — crash #18  (~35 min gap)

Stack trace for each crash:

[openclaw] Uncaught exception: Error: Max reconnect attempts (0) reached after code 1005
    at SafeGatewayPlugin.handleReconnectionAttempt (file:///.../provider-CAlWEl41.js:3318:47)
    at SafeGatewayPlugin.handleClose (file:///.../provider-CAlWEl41.js:3364:8)
    at WebSocket.<anonymous> (file:///.../provider-CAlWEl41.js:3307:9)

The gateway restarts via systemd (~8 seconds), and sessions/channels reconnect. But the Control UI and any in-flight requests are disrupted.

Impact

Gateway process restarts ~50 times per day
Control UI disconnects briefly each time
In-flight agent turns may be interrupted
Logs filled with crash/restart noise
No data loss (sessions persist), but reliability is degraded

Root Cause Analysis

OpenRouter (and likely other providers) periodically cycles WebSocket connections — this is normal behavior
WebSocket code 1005 means "no status received" — the remote end closed without a proper close frame
SafeGatewayPlugin is configured with maxReconnectAttempts: 0 — any connection drop triggers a crash
The plugin does not distinguish between "provider unavailable" and "transient connection drop"

Suggested Fix

Set maxReconnectAttempts to a reasonable value (e.g., 3-5) with exponential backoff
Only crash/restart if reconnection fails after all attempts
Add a configurable reconnectDelay (e.g., 2-5 seconds) before retrying
Optionally add a WebSocket ping/pong keepalive to detect dead connections proactively

Network Verification

Confirmed the issue is NOT related to VPN/Tailscale latency:

$ ip route get 172.67.24.1
172.67.24.1 via 192.168.1.1 dev eno1  ← direct internet, not Tailscale

$ curl -so /dev/null -w "connect: %{time_connect}s, tls: %{time_appconnect}s\n" https://openrouter.ai
connect: 0.007s, tls: 0.038s  ← fast direct connection

The outbound path is direct to the internet with no VPN/proxy in between. The 35-minute interval matches typical load balancer connection timeout behavior.

Logs

Happy to provide full log excerpts or additional debugging if needed.

extent analysis

Fix Plan

To address the issue, we need to modify the SafeGatewayPlugin configuration to handle WebSocket connection drops more robustly. Here are the steps:

Set maxReconnectAttempts to a reasonable value (e.g., 3-5) to allow for retries before crashing.
Implement exponential backoff for reconnection attempts.
Add a configurable reconnectDelay (e.g., 2-5 seconds) before retrying.

Example code changes:

// Update SafeGatewayPlugin configuration
const safeGatewayPlugin = new SafeGatewayPlugin({
  maxReconnectAttempts: 3, // Set to a reasonable value
  reconnectDelay: 2000, // 2 seconds
  backoffFactor: 2, // Exponential backoff factor
});

// Implement exponential backoff for reconnection attempts
function reconnectWithBackoff(attempt) {
  const delay = safeGatewayPlugin.reconnectDelay * Math.pow(safeGatewayPlugin.backoffFactor, attempt);
  setTimeout(() => {
    // Reconnection logic here
  }, delay);
}

Verification

To verify the fix, monitor the gateway logs for crashes and restarts. The number of crashes should decrease significantly. Additionally, check the Control UI and in-flight requests for disruptions.

Run grep "Uncaught exception" openclaw-2026-03-29.log | wc -l to check for crashes.
Verify that the gateway process remains stable and does not restart frequently.
Test in-flight requests and Control UI connectivity to ensure they are not disrupted.

Extra Tips

Consider adding a WebSocket ping/pong keepalive to detect dead connections proactively.
Monitor the gateway logs and adjust the maxReconnectAttempts and reconnectDelay values as needed to achieve a balance between robustness and responsiveness.

Vote matrix · Quick signals

Works

Did the solution work? Tap to confirm.

Easy Fix

Was it a quick fix?

Time Saver

Did it save you time?

Blocking

Was it severely blocking?

Common Issue

Are others likely hitting this too?

Flaky / Intermittent

Is it intermittent?

Verified / Reproducible

Can you reproduce it reliably?

#database connection #vector store #embedding generation #cache error #pipeline error

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Data

Security

Network

Code

UI/UX

Text

System

Multimedia

Protocol

API

Engineering

openclaw - 💡(How to fix) Fix Gateway crashes on provider WebSocket disconnect (code 1005) — no reconnect attempt [2 comments, 1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Code Example

Description

Environment

Evidence

Impact

Root Cause Analysis

Suggested Fix

Network Verification

Logs

extent analysis

Fix Plan

Verification

Extra Tips

Still need to ship something?

TRENDING

openclaw - 💡(How to fix) Fix Gateway crashes on provider WebSocket disconnect (code 1005) — no reconnect attempt [2 comments, 1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Code Example

Description

Environment

Evidence

Impact

Root Cause Analysis

Suggested Fix

Network Verification

Logs

extent analysis

Fix Plan

Verification

Extra Tips

Still need to ship something?

RELATED_DISCOVERY

TRENDING