Forum

Hello,

We have checked the status of your OpenClaw-Zeo service:

• The service is currently in a suspended state, but your disk data is fully preserved: all data including .openclaw/openclaw.json, workspace, identity, ollama, and telegram remains on the /home/node disk and has not been lost.
• The error gateway.bind: gateway.bind host aliases are legacy is thrown by the OpenClaw application itself. This is because the new version of the image you are currently using, ghcr.io/openclaw/openclaw:2026.3.31, has changed the format of the gateway.bind field, making your old configuration file incompatible.
• Consequently, every time the container starts, OpenClaw rejects the configuration file upon reading it, causing the container to exit immediately. (This is not the platform "resetting" your configuration; it is the new version not accepting the old format.)

Since this is a backward-incompatible change at the OpenClaw application layer, the platform cannot directly modify the configuration of third-party applications for you. We recommend that you roll back the image version to a version prior to 2026.3.31 (e.g., ghcr.io/openclaw/openclaw:2026.3.30 or earlier) to get the service running again. Your disk data is still there, and once the service is running, all your workspaces, skills, and login states will be preserved.

Steps:

1. Go to the Zeabur Dashboard → your OpenClaw-Zeo service.
2. Settings → Source → Image, change the tag from 2026.3.31 to 2026.3.30 (or any version you previously ran successfully).
3. Redeploy and restore the service from the suspended state.

After it is running, if you wish to upgrade to the new version, please check the changelog/migration guide on the official OpenClaw GitHub, update the gateway.bind field in openclaw.json according to the new schema, and then proceed with the upgrade.

If you still encounter issues after rolling back, please reply and let us know, and we will assist you further.