openclaw - 💡(How to fix) Fix Silent clean exit (code 0) when OPENCLAW_PLUGIN_STAGE_DIR is on NTFS bind-mounted volume — channels resolve as stopped with no error log [1 comments, 2 participants]

OPENCLAW_PLUGIN_STAGE_DIR was pointing to a path inside a Windows NTFS bind-mounted volume. The plugin staging path performs npm installs, node_modules creation/rename, and symlink operations — all requiring POSIX filesystem behavior. NTFS-backed mounts break this silently, causing startAccount() to resolve cleanly with no exception or log output. When plugin stage-dir filesystem behavior breaks runtime dep loading, OpenClaw should log a loud, descriptive error rather than letting startAccount() resolve cleanly and channels silently become stopped.

TL;DR

Move the OPENCLAW_PLUGIN_STAGE_DIR to a Docker-managed named volume with a Linux filesystem to resolve the silent exit issue.

Guidance

• Verify that the OPENCLAW_PLUGIN_STAGE_DIR is currently pointing to a path inside a Windows NTFS bind-mounted volume, which is causing the issue due to its lack of POSIX filesystem behavior.
• Test the fix by setting OPENCLAW_PLUGIN_STAGE_DIR to a temporary directory like /tmp/fresh-dir to confirm that the issue is resolved.
• Create a Docker-managed named volume with a Linux filesystem and update the OPENCLAW_PLUGIN_STAGE_DIR to point to this new volume.
• Monitor the gateway logs to ensure that the issue is resolved and channels are starting correctly.

Example

No code snippet is necessary in this case, as the fix involves a configuration change rather than a code modification.

Notes

This fix assumes that the issue is caused by the OPENCLAW_PLUGIN_STAGE_DIR being on an NTFS bind-mounted volume, which is not compatible with the required POSIX filesystem behavior. If the issue persists after applying this fix, further investigation may be necessary.

Recommendation

Apply the workaround by moving the OPENCLAW_PLUGIN_STAGE_DIR to a Docker-managed named volume with a Linux filesystem, as this has been confirmed to resolve the issue.