[BUG] Claude Code fails to reconnect after network change (ECONNRESET) — requires process restart

[KindenKuo]

Preflight Checklist

• I have searched existing issues and this hasn't been reported yet
• This is a single bug report (please file separate reports for different bugs)
• I am using the latest version of Claude Code

What's Wrong?

Claude Code fails to reconnect after network change (ECONNRESET) — requires process restart

Bug Description

Claude Code becomes unable to connect to the API after switching between networks (e.g., office Wi-Fi → home Wi-Fi), returning ECONNRESET errors on every attempt. The only workaround is to kill the process (pkill -f claude) and restart it. Without this, even waiting through all 10 retry attempts fails.

Steps to Reproduce

1. Start Claude Code on one network (e.g., office Wi-Fi / Ethernet)
2. Switch to a different network (e.g., go home and connect to home Wi-Fi)
3. Try to use Claude Code (e.g., run /init)

Expected Behavior

Claude Code should detect the network change and establish a new connection to api.anthropic.com.

Actual Behavior

⎿  Unable to connect to API (ECONNRESET)
   Retrying in 7 seconds… (attempt 5/10)

All 10 retry attempts fail with ECONNRESET. The error persists until the Claude Code process is fully killed and restarted.

Diagnosis

The issue appears to be stale HTTP/2 connection pooling. After a network change, the old socket is dead but Claude Code keeps trying to reuse it.

Evidence that the network itself is fine:

• curl -v https://api.anthropic.com — works (HTTP 404 as expected, TLS OK)
• node -e "const https = require('https'); https.get('https://api.anthropic.com', res => console.log(res.statusCode))" — works (returns 404)
• DNS resolution works, no proxy configured, no firewall blocking
• openssl s_client fails with errno=54 (connection reset), suggesting something at the socket/TLS layer

Killing all Claude processes and restarting immediately resolves the issue, confirming it's a stale connection problem rather than a system-level network issue.

Workaround

pkill -f claude
# Then restart Claude Code

Suggested Fix

• Detect ECONNRESET during retries and force-close the existing connection pool before retrying, rather than reusing the same dead socket.
• Optionally, listen for OS-level network change events (e.g., macOS SCNetworkReachability) to proactively reset connections.

Environment

• OS: macOS 26.2
• Claude Code version: v2.1.34
• Network: switching between different Wi-Fi networks
• Node.js: system default

What Should Happen?

Claude should work properly

Error Messages/Logs

⎿  Unable to connect to API (ECONNRESET)

Steps to Reproduce

Start Claude Code on one network (e.g., office Wi-Fi / Ethernet)
Switch to a different network (e.g., go home and connect to home Wi-Fi)
Try to use Claude Code (e.g., run /init)

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

No response

Claude Code Version

Claude Code version: v2.1.34

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

No response

[github-actions]

Found 3 possible duplicate issues:

1. [BUG] Retry mechanism fails to recover from connection loss - stale HTTP/2 connection pool not reset #23081
2. API Connection Errors During Operation in WSL2 (Correlated with IP Changes) #21417
3. Persistent ECONNRESET Errors on macOS Network Connections #5674

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code