TerrifiedBug · TerrifiedBug · Mar 7, 2026 · Mar 7, 2026 · Mar 7, 2026 · Mar 7, 2026
diff --git a/prisma/migrations/20260307000000_add_node_maintenance_mode/migration.sql b/prisma/migrations/20260307000000_add_node_maintenance_mode/migration.sql
@@ -0,0 +1,3 @@
+-- AlterTable
+ALTER TABLE "VectorNode" ADD COLUMN "maintenanceMode" BOOLEAN NOT NULL DEFAULT false;
+ALTER TABLE "VectorNode" ADD COLUMN "maintenanceModeAt" TIMESTAMP(3);
diff --git a/prisma/schema.prisma b/prisma/schema.prisma
@@ -105,6 +105,8 @@ model VectorNode {
   os               String?
   deploymentMode   DeploymentMode       @default(UNKNOWN)
   pendingAction    Json?
+  maintenanceMode  Boolean              @default(false)
+  maintenanceModeAt DateTime?
   pipelineStatuses NodePipelineStatus[]
   nodeMetrics      NodeMetric[]
   pipelineLogs     PipelineLog[]

diff --git a/src/app/(dashboard)/fleet/[nodeId]/page.tsx b/src/app/(dashboard)/fleet/[nodeId]/page.tsx
@@ -3,7 +3,7 @@
 import { useParams, useRouter } from "next/navigation";
 import { useQuery, useMutation, useQueryClient } from "@tanstack/react-query";
 import { useTRPC } from "@/trpc/client";
-import { ArrowLeft, ShieldOff, Trash2, Activity, Terminal, Server, Pencil, Check, X } from "lucide-react";
+import { ArrowLeft, ShieldOff, Trash2, Activity, Terminal, Server, Pencil, Check, X, Wrench } from "lucide-react";
 import { NodeLogs } from "@/components/fleet/node-logs";
 import { toast } from "sonner";
 import { useState } from "react";
@@ -131,6 +131,32 @@ export default function NodeDetailPage() {
     })
   );
 
+  const maintenanceMutation = useMutation(
+    trpc.fleet.setMaintenanceMode.mutationOptions({
+      onSuccess: () => {
+        queryClient.invalidateQueries({ queryKey: trpc.fleet.get.queryKey({ id: params.nodeId }) });
+        queryClient.invalidateQueries({ queryKey: trpc.fleet.list.queryKey() });
+        queryClient.invalidateQueries({ queryKey: trpc.fleet.listWithPipelineStatus.queryKey() });
+      },
+    }),
+  );
+
+  function handleMaintenanceToggle() {
+    if (!node) return;
+    if (!node.maintenanceMode) {
+      const runningCount = node.pipelineStatuses.filter(
+        (s) => s.status === "RUNNING"
+      ).length;
+      if (!confirm(
+        `Enter maintenance mode for "${node.name}"?\n\nThis will stop ${runningCount} running pipeline(s) on this node. Pipelines will automatically resume when maintenance mode is turned off.`
+      )) return;
+    }
+    maintenanceMutation.mutate({
+      nodeId: node.id,
+      enabled: !node.maintenanceMode,
+    });
+  }
+
   function handleRevoke() {
     if (!node) return;
     if (!confirm(`Revoke token for "${node.name}"? The agent will no longer be able to connect.`)) {
@@ -217,6 +243,19 @@ export default function NodeDetailPage() {
           </div>
         </div>
         <div className="flex items-center gap-2">
+          <Button
+            variant={node.maintenanceMode ? "default" : "outline"}
+            size="sm"
+            onClick={handleMaintenanceToggle}
+            disabled={maintenanceMutation.isPending}
+          >
+            <Wrench className="mr-2 h-4 w-4" />
+            {maintenanceMutation.isPending
+              ? "Updating..."
+              : node.maintenanceMode
+                ? "Exit Maintenance"
+                : "Enter Maintenance"}
+          </Button>
           {node.nodeTokenHash && (
             <Button
               variant="outline"
@@ -240,6 +279,28 @@ export default function NodeDetailPage() {
         </div>
       </div>
 
+      {node.maintenanceMode && (
+        <div className="flex items-center gap-3 rounded-lg border border-orange-500/50 bg-orange-50 px-4 py-3 dark:bg-orange-950/20">
+          <Wrench className="h-5 w-5 text-orange-600" />
+          <div className="flex-1">
+            <p className="text-sm font-medium text-orange-800 dark:text-orange-200">
+              This node is in maintenance mode
+            </p>
+            <p className="text-xs text-orange-600 dark:text-orange-400">
+              All pipelines are stopped. They will automatically resume when maintenance mode is turned off.
+            </p>
+          </div>
+          <Button
+            variant="outline"
+            size="sm"
+            onClick={handleMaintenanceToggle}
+            disabled={maintenanceMutation.isPending}
+          >
+            Exit Maintenance
+          </Button>
+        </div>
+      )}
+
       <div>
         {/* Node Details */}
         <Card>

diff --git a/src/app/(dashboard)/fleet/page.tsx b/src/app/(dashboard)/fleet/page.tsx
@@ -23,6 +23,7 @@ import {
   TooltipTrigger,
 } from "@/components/ui/tooltip";
 import { Skeleton } from "@/components/ui/skeleton";
+import { Wrench } from "lucide-react";
 import { DeploymentMatrix } from "@/components/fleet/deployment-matrix";
 import { formatLastSeen } from "@/lib/format";
 import { nodeStatusVariant, nodeStatusLabel } from "@/lib/status";
@@ -87,6 +88,15 @@ export default function FleetPage() {
     }),
   );
 
+  const setMaintenance = useMutation(
+    trpc.fleet.setMaintenanceMode.mutationOptions({
+      onSuccess: () => {
+        queryClient.invalidateQueries({ queryKey: trpc.fleet.list.queryKey() });
+        queryClient.invalidateQueries({ queryKey: trpc.fleet.listWithPipelineStatus.queryKey() });
+      },
+    }),
+  );
+
   return (
     <div className="space-y-6">
       {isLoading ? (
@@ -161,54 +171,83 @@ export default function FleetPage() {
                   </div>
                 </TableCell>
                 <TableCell>
-                  <StatusBadge variant={nodeStatusVariant(node.status)}>
-                    {nodeStatusLabel(node.status)}
-                  </StatusBadge>
+                  {node.maintenanceMode ? (
+                    <Badge variant="outline" className="text-orange-600 border-orange-500/50">
+                      <Wrench className="mr-1 h-3 w-3" />
+                      Maintenance
+                    </Badge>
+                  ) : (
+                    <StatusBadge variant={nodeStatusVariant(node.status)}>
+                      {nodeStatusLabel(node.status)}
+                    </StatusBadge>
+                  )}
                 </TableCell>
                 <TableCell className="text-muted-foreground">
                   {formatLastSeen(node.lastSeen)}
                 </TableCell>
                 <TableCell>
-                  {node.pendingAction ? (
-                    <Badge variant="outline" className="text-blue-600">
-                      Update pending...
-                    </Badge>
-                  ) : node.deploymentMode === "DOCKER" ? (
-                    getNodeLatest(node).version &&
-                    node.agentVersion &&
-                    isVersionOlder(node.agentVersion, getNodeLatest(node).version ?? "") ? (
-                      <Tooltip>
-                        <TooltipTrigger asChild>
-                          <span>
-                            <Button variant="outline" size="sm" disabled>
-                              Update
-                            </Button>
-                          </span>
-                        </TooltipTrigger>
-                        <TooltipContent>Update via Docker image pull</TooltipContent>
-                      </Tooltip>
-                    ) : null
-                  ) : getNodeLatest(node).version &&
-                    node.agentVersion &&
-                    isVersionOlder(node.agentVersion, getNodeLatest(node).version ?? "") ? (
+                  <div className="flex items-center gap-2">
                     <Button
-                      variant="outline"
+                      variant={node.maintenanceMode ? "default" : "outline"}
                       size="sm"
-                      disabled={triggerUpdate.isPending}
+                      disabled={setMaintenance.isPending && setMaintenance.variables?.nodeId === node.id}
                       onClick={(e) => {
                         e.preventDefault();
-                        const latest = getNodeLatest(node);
-                        triggerUpdate.mutate({
+                        if (!node.maintenanceMode) {
+                          if (!confirm(
+                            `Enter maintenance mode for "${node.name}"?\n\nThis will stop all running pipelines on this node. Pipelines will automatically resume when maintenance mode is turned off.`
+                          )) return;
+                        }
+                        setMaintenance.mutate({
                           nodeId: node.id,
-                          targetVersion: latest.version!,
-                          downloadUrl: `https://github.com/${AGENT_REPO}/releases/download/${latest.tag}/vf-agent-linux-amd64`,
-                          checksum: `sha256:${latest.checksums["vf-agent-linux-amd64"] ?? ""}`,
+                          enabled: !node.maintenanceMode,
                         });
                       }}
                     >
-                      {triggerUpdate.isPending ? "Updating..." : "Update"}
+                      <Wrench className="mr-1 h-3.5 w-3.5" />
+                      {node.maintenanceMode ? "Exit Maintenance" : "Maintenance"}
                     </Button>
-                  ) : null}
+                    {node.pendingAction ? (
+                      <Badge variant="outline" className="text-blue-600">
+                        Update pending...
+                      </Badge>
+                    ) : node.deploymentMode === "DOCKER" ? (
+                      getNodeLatest(node).version &&
+                      node.agentVersion &&
+                      isVersionOlder(node.agentVersion, getNodeLatest(node).version ?? "") ? (
+                        <Tooltip>
+                          <TooltipTrigger asChild>
+                            <span>
+                              <Button variant="outline" size="sm" disabled>
+                                Update
+                              </Button>
+                            </span>
+                          </TooltipTrigger>
+                          <TooltipContent>Update via Docker image pull</TooltipContent>
+                        </Tooltip>
+                      ) : null
+                    ) : getNodeLatest(node).version &&
+                      node.agentVersion &&
+                      isVersionOlder(node.agentVersion, getNodeLatest(node).version ?? "") ? (
+                      <Button
+                        variant="outline"
+                        size="sm"
+                        disabled={triggerUpdate.isPending}
+                        onClick={(e) => {
+                          e.preventDefault();
+                          const latest = getNodeLatest(node);
+                          triggerUpdate.mutate({
+                            nodeId: node.id,
+                            targetVersion: latest.version!,
+                            downloadUrl: `https://github.com/${AGENT_REPO}/releases/download/${latest.tag}/vf-agent-linux-amd64`,
+                            checksum: `sha256:${latest.checksums["vf-agent-linux-amd64"] ?? ""}`,
+                          });
+                        }}
+                      >
+                        {triggerUpdate.isPending ? "Updating..." : "Update"}
+                      </Button>
+                    ) : null}
+                  </div>
                 </TableCell>
               </TableRow>
             ))}

diff --git a/src/app/api/agent/config/route.ts b/src/app/api/agent/config/route.ts
@@ -16,9 +16,26 @@ export async function GET(request: Request) {
     // Fetch the node to check for pending actions (e.g., self-update)
     const node = await prisma.vectorNode.findUnique({
       where: { id: agent.nodeId },
-      select: { pendingAction: true },
+      select: { pendingAction: true, maintenanceMode: true },
     });
 
+    if (node?.maintenanceMode) {
+      const environment = await prisma.environment.findUnique({
+        where: { id: agent.environmentId },
+        select: { secretBackend: true },
+      });
+      const settings = await prisma.systemSettings.findUnique({
+        where: { id: "singleton" },
+        select: { fleetPollIntervalMs: true },
+      });
+      return NextResponse.json({
+        pipelines: [],
+        pollIntervalMs: settings?.fleetPollIntervalMs ?? 15_000,
+        secretBackend: environment?.secretBackend ?? "BUILTIN",
+        pendingAction: node.pendingAction ?? undefined,
+      });
+    }
+
     const environment = await prisma.environment.findUnique({
       where: { id: agent.environmentId },
       select: {

diff --git a/src/components/fleet/deployment-matrix.tsx b/src/components/fleet/deployment-matrix.tsx
@@ -3,7 +3,7 @@
 import { useQuery } from "@tanstack/react-query";
 import { useTRPC } from "@/trpc/client";
 import { Badge } from "@/components/ui/badge";
-import { Minus } from "lucide-react";
+import { Minus, Wrench } from "lucide-react";
 import Link from "next/link";
 import { StatusDot } from "@/components/ui/status-dot";
 import { pipelineStatusVariant, pipelineStatusLabel } from "@/lib/status";
@@ -51,10 +51,18 @@ export function DeploymentMatrix({ environmentId }: DeploymentMatrixProps) {
             {nodes.map((node) => (
               <th
                 key={node.id}
-                className="px-3 py-2 text-center font-medium text-muted-foreground"
+                className={`px-3 py-2 text-center font-medium text-muted-foreground ${
+                  node.maintenanceMode ? "bg-orange-50/50 dark:bg-orange-950/10" : ""
+                }`}
               >
                 <div>{node.name}</div>
                 <div className="text-xs font-normal">{node.host}</div>
+                {node.maintenanceMode && (
+                  <div className="mt-1 flex items-center justify-center gap-1 text-xs text-orange-600 dark:text-orange-400">
+                    <Wrench className="h-3 w-3" />
+                    Maintenance
+                  </div>
+                )}
               </th>
             ))}
           </tr>
@@ -79,7 +87,7 @@ export function DeploymentMatrix({ environmentId }: DeploymentMatrixProps) {
 
                 if (!ps) {
                   return (
-                    <td key={node.id} className="px-3 py-2 text-center">
+                    <td key={node.id} className={`px-3 py-2 text-center ${node.maintenanceMode ? "opacity-30" : ""}`}>
                       <div className="flex items-center justify-center">
                         <Minus className="h-4 w-4 text-muted-foreground/50" />
                       </div>
@@ -90,7 +98,7 @@ export function DeploymentMatrix({ environmentId }: DeploymentMatrixProps) {
                 const isOutdated = ps.version < pipeline.latestVersion;
 
                 return (
-                  <td key={node.id} className="px-3 py-2 text-center">
+                  <td key={node.id} className={`px-3 py-2 text-center ${node.maintenanceMode ? "opacity-30" : ""}`}>
                     <div className="flex flex-col items-center gap-0.5">
                       {isOutdated ? (
                         <div

diff --git a/src/server/routers/fleet.ts b/src/server/routers/fleet.ts
@@ -264,6 +264,31 @@ export const fleetRouter = router({
       });
     }),
 
+  setMaintenanceMode: protectedProcedure
+    .input(
+      z.object({
+        nodeId: z.string(),
+        enabled: z.boolean(),
+      }),
+    )
+    .use(withTeamAccess("ADMIN"))
+    .use(withAudit("node.maintenance_toggled", "VectorNode"))
+    .mutation(async ({ input }) => {
+      const node = await prisma.vectorNode.findUnique({
+        where: { id: input.nodeId },
+      });
+      if (!node) {
+        throw new TRPCError({ code: "NOT_FOUND", message: "Node not found" });
+      }
+      return prisma.vectorNode.update({
+        where: { id: input.nodeId },
+        data: {
+          maintenanceMode: input.enabled,
+          maintenanceModeAt: input.enabled ? new Date() : null,
+        },
+      });
+    }),
+
   listWithPipelineStatus: protectedProcedure
     .input(z.object({ environmentId: z.string() }))
     .use(withTeamAccess("VIEWER"))