fix(server): avoid duplicate monitor recovery wakes

Co-Authored-By: Paperclip <noreply@paperclip.ing>

server/src/__tests__/heartbeat-process-recovery.test.ts

@@ -1377,6 +1377,32 @@ describeEmbeddedPostgres("heartbeat orphaned process recovery", () => {
     });
   });
 
+  it("does not retry a lost monitor dispatch while another monitor wake remains scheduled", async () => {
+    const { companyId, runId, issueId } = await seedRunFixture({
+      adapterType: "openclaw_gateway",
+      agentStatus: "idle",
+      processPid: null,
+      processGroupId: null,
+      contextSnapshot: {
+        wakeReason: "issue_monitor_due",
+      },
+    });
+    await db
+      .update(issues)
+      .set({ monitorNextCheckAt: new Date("2026-03-19T00:00:00.000Z") })
+      .where(and(eq(issues.id, issueId), eq(issues.companyId, companyId)));
+
+    const heartbeat = heartbeatService(db);
+    const result = await heartbeat.reapOrphanedRuns();
+
+    expect(result).toEqual({ reaped: 1, runIds: [runId] });
+    const retries = await db
+      .select()
+      .from(heartbeatRuns)
+      .where(and(eq(heartbeatRuns.companyId, companyId), eq(heartbeatRuns.retryOfRunId, runId)));
+    expect(retries).toHaveLength(0);
+  });
+
   it("interrupts running runs on graceful shutdown and queues restart recovery without recording a failure", async () => {
     const { agentId, runId, issueId, wakeupRequestId } = await seedRunFixture({
       agentStatus: "running",

server/src/services/heartbeat.ts

@@ -10923,6 +10923,29 @@ export function heartbeatService(db: Db, options: HeartbeatServiceOptions = {})
       .innerJoin(agents, eq(heartbeatRuns.agentId, agents.id))
       .where(eq(heartbeatRuns.status, "running"));
 
+    const monitorIssueIds = [...new Set(activeRuns.flatMap(({ run }) => {
+      const runContext = parseObject(run.contextSnapshot);
+      if (readNonEmptyString(runContext.wakeReason) !== "issue_monitor_due") return [];
+      const issueId = readNonEmptyString(runContext.issueId);
+      return issueId ? [issueId] : [];
+    }))];
+    const monitorIssues = monitorIssueIds.length > 0
+      ? await db
+        .select({
+          id: issues.id,
+          companyId: issues.companyId,
+          monitorNextCheckAt: issues.monitorNextCheckAt,
+        })
+        .from(issues)
+        .where(inArray(issues.id, monitorIssueIds))
+      : [];
+    const monitorNextCheckAtByIssue = new Map(
+      monitorIssues.map((issue) => [
+        `${issue.companyId}:${issue.id}`,
+        issue.monitorNextCheckAt,
+      ]),
+    );
+
     const reaped: string[] = [];
 
     for (const { run, adapterType, adapterConfig } of activeRuns) {
@@ -10973,17 +10996,7 @@ export function heartbeatService(db: Db, options: HeartbeatServiceOptions = {})
       const monitorDispatchLostWithoutFutureWake =
         readNonEmptyString(runContext.wakeReason) === "issue_monitor_due" &&
         Boolean(monitorIssueId) &&
-        await db
-          .select({ monitorNextCheckAt: issues.monitorNextCheckAt })
-          .from(issues)
-          .where(and(eq(issues.id, monitorIssueId!), eq(issues.companyId, run.companyId)))
-          .limit(1)
-          .then((rows) => {
-            const issue = rows[0];
-            if (!issue) return false;
-            const nextCheckAt = issue.monitorNextCheckAt;
-            return !nextCheckAt || nextCheckAt.getTime() <= now.getTime();
-          });
+        monitorNextCheckAtByIssue.get(`${run.companyId}:${monitorIssueId}`) === null;
       const shouldRetry = (run.processLossRetryCount ?? 0) < 1 && (
         (tracksLocalChild && (!!run.processPid || !!run.processGroupId)) ||
         monitorDispatchLostWithoutFutureWake