fix(config): recover critical config clobbers

steipete · steipete · commit 7b2c9a6fa3d3 · 2026-04-22T23:35:48.000+01:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -23,6 +23,7 @@ Docs: https://docs.openclaw.ai
 
 - Config/models: merge provider-scoped model allowlist updates and protect model/provider map writes from accidental full replacement, adding `config set --merge` for additive updates and `--replace` for intentional clobbers. Fixes #65920, #68392, and #68653.
 - Agents/Pi auth: preserve AWS SDK-authenticated Bedrock runs for IMDS and task-role setups, clear stale refresh timers on sentinel fallback, and log unexpected runtime-auth prep failures instead of silently leaving the provider unauthenticated. Thanks @wirjo.
+- Config/gateway: restore last-known-good config on critical clobber signatures such as missing metadata, missing `gateway.mode`, or sharp size drops, preventing gateway crash loops when a valid backup exists. Fixes #70336.
 - Config/gateway: recover configs accidentally prefixed with non-JSON output during gateway startup or `openclaw doctor --fix`, preserving the clobbered file as a backup while leaving normal config reads read-only.
 - Agents/GitHub Copilot: normalize connection-bound Responses item IDs in the Copilot provider wrapper so replayed histories no longer fail after the upstream connection changes. (#69362) Thanks @Menci.
 - Pi embedded runs: pass real built-in tools into Pi session creation and then narrow active tool names after custom tool registration, so the runner and compaction paths compile cleanly and keep OpenClaw-managed custom tool allowlists without feeding string arrays into `createAgentSession`. Thanks @vincentkoc.
diff --git a/docs/gateway/configuration.md b/docs/gateway/configuration.md
@@ -100,6 +100,9 @@ The Gateway also keeps a trusted last-known-good copy after a successful startup
 `openclaw.json` is later changed outside OpenClaw and no longer validates, startup
 and hot reload preserve the broken file as a timestamped `.clobbered.*` snapshot,
 restore the last-known-good copy, and log a loud warning with the recovery reason.
+Startup read recovery also treats sharp size drops, missing config metadata, and a
+missing `gateway.mode` as critical clobber signatures when the last-known-good
+copy had those fields.
 If a status/log line is accidentally prepended before an otherwise valid JSON
 config, gateway startup and `openclaw doctor --fix` can strip the prefix,
 preserve the polluted file as `.clobbered.*`, and continue with the recovered
diff --git a/docs/gateway/troubleshooting.md b/docs/gateway/troubleshooting.md
@@ -303,6 +303,7 @@ Common signatures:
 - `.clobbered.*` exists → an external direct edit or startup read was restored.
 - `.rejected.*` exists → an OpenClaw-owned config write failed schema or clobber checks before commit.
 - `Config write rejected:` → the write tried to drop required shape, shrink the file sharply, or persist invalid config.
+- `missing-meta-vs-last-good`, `gateway-mode-missing-vs-last-good`, or `size-drop-vs-last-good:*` → startup treated the current file as clobbered because it lost fields or size compared with the last-known-good backup.
 - `Config last-known-good promotion skipped` → the candidate contained redacted secret placeholders such as `***`.
 
 Fix options:
diff --git a/src/config/io.observe-recovery.test.ts b/src/config/io.observe-recovery.test.ts
@@ -20,6 +20,7 @@ describe("config observe recovery", () => {
   const clobberedUpdateChannelConfig = { update: { channel: "beta" } };
   const clobberedUpdateChannelRaw = `${JSON.stringify(clobberedUpdateChannelConfig, null, 2)}\n`;
   const recoverableTelegramConfig = {
+    meta: { lastTouchedAt: "2026-04-22T00:00:00.000Z" },
     update: { channel: "beta" },
     gateway: { mode: "local" },
     channels: { telegram: { enabled: true, dmPolicy: "pairing", groupPolicy: "allowlist" } },
@@ -49,6 +50,12 @@ describe("config observe recovery", () => {
     await fsp.copyFile(configPath, `${configPath}.bak`);
   }
 
+  async function writeConfigRaw(configPath: string, config: Record<string, unknown>) {
+    const raw = `${JSON.stringify(config, null, 2)}\n`;
+    await fsp.writeFile(configPath, raw, "utf-8");
+    return { raw, parsed: config };
+  }
+
   async function writeClobberedUpdateChannel(configPath: string) {
     await fsp.writeFile(configPath, clobberedUpdateChannelRaw, "utf-8");
     return {
@@ -82,6 +89,20 @@ describe("config observe recovery", () => {
     });
   }
 
+  async function recoverSuspiciousConfigRead(params: {
+    deps: ObserveRecoveryDeps;
+    configPath: string;
+    raw: string;
+    parsed: unknown;
+  }) {
+    return await maybeRecoverSuspiciousConfigRead({
+      deps: params.deps,
+      configPath: params.configPath,
+      raw: params.raw,
+      parsed: params.parsed,
+    });
+  }
+
   function recoverClobberedUpdateChannelSync(params: {
     deps: ObserveRecoveryDeps;
     configPath: string;
@@ -142,6 +163,7 @@ describe("config observe recovery", () => {
     await withSuiteHome(async (home) => {
       const { deps, configPath, auditPath, warn } = makeDeps(home);
       await seedConfigBackup(configPath, {
+        meta: { lastTouchedAt: "2026-04-22T00:00:00.000Z" },
         update: { channel: "beta" },
         browser: { enabled: true },
         gateway: { mode: "local", auth: { mode: "token", token: "secret-token" } },
@@ -165,6 +187,97 @@ describe("config observe recovery", () => {
     });
   });
 
+  it("auto-restores when metadata disappears from an otherwise valid config", async () => {
+    await withSuiteHome(async (home) => {
+      const { deps, configPath, auditPath } = makeDeps(home);
+      await seedConfigBackup(configPath, recoverableTelegramConfig);
+      const clobbered = await writeConfigRaw(configPath, {
+        update: { channel: "beta" },
+        gateway: { mode: "local" },
+        channels: { telegram: { enabled: true, dmPolicy: "pairing", groupPolicy: "allowlist" } },
+      });
+
+      const recovered = await recoverSuspiciousConfigRead({ deps, configPath, ...clobbered });
+
+      expect((recovered.parsed as { meta?: unknown }).meta).toEqual(recoverableTelegramConfig.meta);
+      const observe = await readLastObserveEvent(auditPath);
+      expect(observe?.restoredFromBackup).toBe(true);
+      expect(observe?.suspicious).toEqual(expect.arrayContaining(["missing-meta-vs-last-good"]));
+    });
+  });
+
+  it("auto-restores when gateway mode disappears from the last-good shape", async () => {
+    await withSuiteHome(async (home) => {
+      const { deps, configPath, auditPath } = makeDeps(home);
+      await seedConfigBackup(configPath, recoverableTelegramConfig);
+      const clobbered = await writeConfigRaw(configPath, {
+        meta: { lastTouchedAt: "2026-04-22T00:00:00.000Z" },
+        update: { channel: "beta" },
+        channels: { telegram: { enabled: true, dmPolicy: "pairing", groupPolicy: "allowlist" } },
+      });
+
+      const recovered = await recoverSuspiciousConfigRead({ deps, configPath, ...clobbered });
+
+      expect((recovered.parsed as { gateway?: { mode?: string } }).gateway?.mode).toBe("local");
+      const observe = await readLastObserveEvent(auditPath);
+      expect(observe?.restoredFromBackup).toBe(true);
+      expect(observe?.suspicious).toEqual(
+        expect.arrayContaining(["gateway-mode-missing-vs-last-good"]),
+      );
+    });
+  });
+
+  it("auto-restores after a large size drop against last-good config", async () => {
+    await withSuiteHome(async (home) => {
+      const { deps, configPath, auditPath } = makeDeps(home);
+      await seedConfigBackup(configPath, {
+        ...recoverableTelegramConfig,
+        channels: {
+          telegram: {
+            enabled: true,
+            dmPolicy: "pairing",
+            groupPolicy: "allowlist",
+            allowFrom: Array.from({ length: 60 }, (_, index) => `telegram-user-${index}`),
+          },
+        },
+      });
+      const clobbered = await writeConfigRaw(configPath, {
+        meta: { lastTouchedAt: "2026-04-22T00:00:00.000Z" },
+        gateway: { mode: "local" },
+      });
+
+      const recovered = await recoverSuspiciousConfigRead({ deps, configPath, ...clobbered });
+
+      expect(
+        (recovered.parsed as { channels?: { telegram?: { allowFrom?: string[] } } }).channels
+          ?.telegram?.allowFrom,
+      ).toHaveLength(60);
+      const observe = await readLastObserveEvent(auditPath);
+      expect(observe?.restoredFromBackup).toBe(true);
+      expect(observe?.suspicious).toEqual(
+        expect.arrayContaining([expect.stringMatching(/^size-drop-vs-last-good:/)]),
+      );
+    });
+  });
+
+  it("does not restore noncritical config edits", async () => {
+    await withSuiteHome(async (home) => {
+      const { deps, configPath, auditPath } = makeDeps(home);
+      await seedConfigBackup(configPath, recoverableTelegramConfig);
+      const editedConfig = {
+        ...recoverableTelegramConfig,
+        update: { channel: "stable" },
+      };
+      const edited = await writeConfigRaw(configPath, editedConfig);
+
+      const recovered = await recoverSuspiciousConfigRead({ deps, configPath, ...edited });
+
+      expect(recovered.parsed).toEqual(editedConfig);
+      await expect(fsp.readFile(configPath, "utf-8")).resolves.toBe(edited.raw);
+      await expect(fsp.stat(auditPath)).rejects.toThrow();
+    });
+  });
+
   it("dedupes repeated suspicious hashes", async () => {
     await withSuiteHome(async (home) => {
       const { deps, configPath, auditPath } = makeDeps(home);
diff --git a/src/config/io.observe-recovery.ts b/src/config/io.observe-recovery.ts
@@ -441,6 +441,15 @@ function resolveSuspiciousSignature(
   return `${current.hash}:${suspicious.join(",")}`;
 }
 
+function isRecoverableConfigReadSuspiciousReason(reason: string): boolean {
+  return (
+    reason === "missing-meta-vs-last-good" ||
+    reason === "gateway-mode-missing-vs-last-good" ||
+    reason === "update-channel-only-root" ||
+    reason.startsWith("size-drop-vs-last-good:")
+  );
+}
+
 function resolveConfigReadRecoveryContext(params: {
   current: ConfigHealthFingerprint;
   parsed: unknown;
@@ -454,7 +463,7 @@ function resolveConfigReadRecoveryContext(params: {
     parsed: params.parsed,
     lastKnownGood: params.backupBaseline,
   });
-  if (!suspicious.includes("update-channel-only-root")) {
+  if (!suspicious.some(isRecoverableConfigReadSuspiciousReason)) {
     return null;
   }
   const suspiciousSignature = resolveSuspiciousSignature(params.current, suspicious);
