openclaw - 💡(How to fix) Fix [Bug]: placeholder.openclaw.cloud unreachable on WSL2 with custom gateway port [1 comments, 2 participants]

Bug type

Behavior bug (incorrect output/state without crash)

Beta release blocker

No

Summary

When uploading an image through Open WebUI → OpenClaw → lmstudio, the vision pipeline fails because placeholder.openclaw.cloud is unreachable. The image is received by OpenClaw successfully, but when the vision model tries to fetch it from the internal media URL, the request fails.

Root cause analysis

placeholder.openclaw.cloud appears to be a runtime-generated internal media hosting domain. Two issues:

1. DNS: The domain is not resolvable in the local environment — it's not defined in OpenClaw's source code (grep found no references), suggesting it's either dynamically generated or expected to be provided by an external service that isn't running.

2. Port mismatch: Even after adding a hosts entry pointing to 127.0.0.1, the request fails because:

   • The URL uses default port (80/443)
   • Gateway is listening on 127.0.0.1:18789
   • No service is listening on port 80

3. Private IP blocking: When the hosts entry resolves to 127.0.0.1, OpenClaw's security mechanism may also block the request as a private IP address.

Workaround attempted

• Added 127.0.0.1 placeholder.openclaw.cloud to /etc/hosts — resolves DNS but still fails due to port mismatch and private IP blocking
• Confirmed vision works when bypassing OpenClaw (Open WebUI → lmstudio directly)
• Confirmed vision works when calling lmstudio directly

Steps to reproduce

1. User uploads image via Open WebUI
2. OpenClaw receives the image binary and stores it internally
3. OpenClaw generates a media URL: https://placeholder.openclaw.cloud/<id>/...
4. Vision pipeline attempts to fetch the image from this URL
5. Request fails:
   • Without hosts entry: ENOTFOUND (DNS resolution failure)
   • With 127.0.0.1 placeholder.openclaw.cloud in /etc/hosts: ECONNREFUSED or blocked by private IP security check

Expected behavior

The vision pipeline should be able to access images that were uploaded through the chat, regardless of whether the gateway is running on a non-default port or in a WSL2 environment.

Actual behavior

When uploading an image through Open WebUI → OpenClaw → lmstudio, the vision pipeline fails because placeholder.openclaw.cloud is unreachable. The image is received by OpenClaw successfully, but when the vision model tries to fetch it from the internal media URL, the request fails.

OpenClaw version

2026.5.7

Operating system

Windows 11 WSL2 (Ubuntu)

Install method

No response

Model

Qwen 3.6-27B

Provider / routing chain

Open WebUI (WSL docker/mirrowed)→ OpenClaw (WSL)→ lmstudio/qwen3.6:27b (windows)

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response