Adapted community discussion: Paraphrased and condensed from an anonymized public OpenClaw Discord support conversation from August 2026. Technical guidance was reviewed for OpenClaw 2026.9.3.
A routine update failed, and I made things worse by killing processes and reinstalling over the existing setup. A partial SQLite migration ran, the Gateway will not start, and I am worried about losing agents, channels, cron jobs, and workspace data. Where should recovery begin?
The current error is more specific: multiple agents are configured, but the terminal operation has no explicit owner. The CLI asks for an agent ID instead of starting normally.
That means the config is readable and OpenClaw refuses to guess which agent owns the operation. List the real IDs and bindings first:
openclaw agents list --bindings
Then pass --agent <id> to an agent-scoped command, or open the terminal UI with a session key for that agent. Do not delete agents or assign a default until you identify which one owns the existing sessions and channel bindings.
I eventually rebuilt enough of the config to bring the main agent back, but some settings were lost and the channels and cron jobs still fail. Repeated update and Doctor loops made the recovery harder to reason about.
Stop rerunning the update and Doctor now that the core agent is back. Do not recreate or manually execute scheduled jobs until their ownership and state are known, because that can duplicate real work. Collect openclaw agents list --bindings, openclaw cron status, openclaw cron list --json, and openclaw gateway status --deep --json. Treat a non-converging migration as an upgrade regression and preserve the evidence for a report.