Add auto-pause TTL and ping endpoint for sandbox inactivity management

Replace the existing auto-destroy TTL behavior with auto-pause: when a
sandbox exceeds its timeout_sec of inactivity, the TTL reaper now pauses
it (snapshot + teardown) instead of destroying it, preserving the ability
to resume later.

Key changes:
- TTL reaper calls Pause instead of Destroy, with fallback to Destroy if
  pause fails (e.g. Firecracker process already gone)
- New PingSandbox RPC resets the in-memory LastActiveAt timer
- New POST /v1/sandboxes/{id}/ping REST endpoint resets both agent memory
  and DB last_active_at
- ListSandboxes RPC now includes auto_paused_sandbox_ids so the reconciler
  can distinguish auto-paused sandboxes from crashed ones in a single call
- Reconciler polls every 5s (was 30s) and marks auto-paused as "paused"
  vs orphaned as "stopped"
- Resume RPC accepts timeout_sec from DB so TTL survives pause/resume cycles
- Reaper checks every 2s (was 10s) and uses a detached context to avoid
  incomplete pauses on app shutdown
- Default timeout_sec changed from 300 to 0 (no auto-pause unless requested)

cmd/control-plane/main.go

@@ -59,7 +59,7 @@ func main() {
 	srv := api.New(queries, agentClient, pool, []byte(cfg.JWTSecret))
 
 	// Start reconciler.
-	reconciler := api.NewReconciler(queries, agentClient, "default", 30*time.Second)
+	reconciler := api.NewReconciler(queries, agentClient, "default", 5*time.Second)
 	reconciler.Start(ctx)
 
 	httpServer := &http.Server{

db/migrations/20260310094104_initial.sql

@@ -8,7 +8,7 @@ CREATE TABLE sandboxes (
     status          TEXT NOT NULL DEFAULT 'pending',
     vcpus           INTEGER NOT NULL DEFAULT 1,
     memory_mb       INTEGER NOT NULL DEFAULT 512,
-    timeout_sec     INTEGER NOT NULL DEFAULT 300,
+    timeout_sec     INTEGER NOT NULL DEFAULT 0,
     guest_ip        TEXT NOT NULL DEFAULT '',
     host_ip         TEXT NOT NULL DEFAULT '',
     created_at      TIMESTAMPTZ NOT NULL DEFAULT NOW(),

go.mod

@@ -9,6 +9,7 @@ require (
 	github.com/google/uuid v1.6.0
 	github.com/gorilla/websocket v1.5.3
 	github.com/jackc/pgx/v5 v5.8.0
+	github.com/joho/godotenv v1.5.1
 	github.com/vishvananda/netlink v1.1.1-0.20210330154013-f5de75959ad5
 	github.com/vishvananda/netns v0.0.0-20210104183010-2eb08e3e575f
 	golang.org/x/crypto v0.49.0
@@ -20,7 +21,6 @@ require (
 	github.com/jackc/pgpassfile v1.0.0 // indirect
 	github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
 	github.com/jackc/puddle/v2 v2.2.2 // indirect
-	github.com/joho/godotenv v1.5.1 // indirect
 	golang.org/x/sync v0.20.0 // indirect
 	golang.org/x/text v0.35.0 // indirect
 )

internal/api/handlers_sandbox.go

@@ -99,9 +99,7 @@ func (h *sandboxHandler) Create(w http.ResponseWriter, r *http.Request) {
 	if req.MemoryMB <= 0 {
 		req.MemoryMB = 512
 	}
-	if req.TimeoutSec <= 0 {
-		req.TimeoutSec = 300
-	}
+	// timeout_sec = 0 means no auto-pause; only set if explicitly requested.
 
 	ctx := r.Context()
 	ac := auth.MustFromContext(ctx)
@@ -259,7 +257,8 @@ func (h *sandboxHandler) Resume(w http.ResponseWriter, r *http.Request) {
 	}
 
 	resp, err := h.agent.ResumeSandbox(ctx, connect.NewRequest(&pb.ResumeSandboxRequest{
-		SandboxId: sandboxID,
+		SandboxId:  sandboxID,
+		TimeoutSec: sb.TimeoutSec,
 	}))
 	if err != nil {
 		status, code, msg := agentErrToHTTP(err)
@@ -285,6 +284,44 @@ func (h *sandboxHandler) Resume(w http.ResponseWriter, r *http.Request) {
 	writeJSON(w, http.StatusOK, sandboxToResponse(sb))
 }
 
+// Ping handles POST /v1/sandboxes/{id}/ping.
+// Resets the inactivity timer for a running sandbox.
+func (h *sandboxHandler) Ping(w http.ResponseWriter, r *http.Request) {
+	sandboxID := chi.URLParam(r, "id")
+	ctx := r.Context()
+	ac := auth.MustFromContext(ctx)
+
+	sb, err := h.db.GetSandboxByTeam(ctx, db.GetSandboxByTeamParams{ID: sandboxID, TeamID: ac.TeamID})
+	if err != nil {
+		writeError(w, http.StatusNotFound, "not_found", "sandbox not found")
+		return
+	}
+	if sb.Status != "running" {
+		writeError(w, http.StatusConflict, "invalid_state", "sandbox is not running")
+		return
+	}
+
+	if _, err := h.agent.PingSandbox(ctx, connect.NewRequest(&pb.PingSandboxRequest{
+		SandboxId: sandboxID,
+	})); err != nil {
+		status, code, msg := agentErrToHTTP(err)
+		writeError(w, status, code, msg)
+		return
+	}
+
+	if err := h.db.UpdateLastActive(ctx, db.UpdateLastActiveParams{
+		ID: sandboxID,
+		LastActiveAt: pgtype.Timestamptz{
+			Time:  time.Now(),
+			Valid: true,
+		},
+	}); err != nil {
+		slog.Warn("ping: failed to update last_active_at in DB", "sandbox_id", sandboxID, "error", err)
+	}
+
+	w.WriteHeader(http.StatusNoContent)
+}
+
 // Destroy handles DELETE /v1/sandboxes/{id}.
 func (h *sandboxHandler) Destroy(w http.ResponseWriter, r *http.Request) {
 	sandboxID := chi.URLParam(r, "id")

internal/api/handlers_test_ui.go

@@ -175,8 +175,8 @@ const testUIHTML = `<!DOCTYPE html>
     <input type="number" id="create-vcpus" value="1" min="1" max="8">
     <label>Memory (MB)</label>
     <input type="number" id="create-memory" value="512" min="128" max="8192">
-    <label>Timeout (sec)</label>
-    <input type="number" id="create-timeout" value="300" min="30">
+    <label>Timeout (sec, 0 = no auto-pause)</label>
+    <input type="number" id="create-timeout" value="0" min="0">
     <div class="btn-row">
       <button class="btn-green" onclick="createSandbox()">Create</button>
     </div>
@@ -417,7 +417,7 @@ function renderSandboxes(sandboxes) {
     document.getElementById('sandboxes-table').innerHTML = '<p style="color:#5f5c57;margin-top:8px">No sandboxes</p>';
     return;
   }
-  let html = '<table><thead><tr><th>ID</th><th>Status</th><th>Template</th><th>vCPUs</th><th>Mem</th><th>Host IP</th><th>Created</th><th>Actions</th></tr></thead><tbody>';
+  let html = '<table><thead><tr><th>ID</th><th>Status</th><th>Template</th><th>vCPUs</th><th>Mem</th><th>TTL</th><th>Host IP</th><th>Created</th><th>Actions</th></tr></thead><tbody>';
   for (const sb of sandboxes) {
     html += '<tr>';
     html += '<td class="clickable" onclick="useSandbox(\'' + sb.id + '\')">' + sb.id + '</td>';
@@ -425,10 +425,12 @@ function renderSandboxes(sandboxes) {
     html += '<td>' + esc(sb.template) + '</td>';
     html += '<td>' + sb.vcpus + '</td>';
     html += '<td>' + sb.memory_mb + 'MB</td>';
+    html += '<td>' + (sb.timeout_sec ? sb.timeout_sec + 's' : '-') + '</td>';
     html += '<td>' + (sb.host_ip || '-') + '</td>';
     html += '<td>' + new Date(sb.created_at).toLocaleTimeString() + '</td>';
     html += '<td><div class="btn-row">';
     if (sb.status === 'running') {
+      html += '<button class="btn-blue" onclick="pingSandbox(\'' + sb.id + '\')">Ping</button>';
       html += '<button class="btn-amber" onclick="pauseSandbox(\'' + sb.id + '\')">Pause</button>';
       html += '<button class="btn-red" onclick="destroySandbox(\'' + sb.id + '\')">Destroy</button>';
     } else if (sb.status === 'paused') {
@@ -497,6 +499,16 @@ async function destroySandbox(id) {
   }
 }
 
+async function pingSandbox(id) {
+  log('Pinging ' + id + '...', 'info');
+  try {
+    await api('POST', '/v1/sandboxes/' + id + '/ping', null, 'apikey');
+    log('Pinged ' + id + ' — inactivity timer reset', 'ok');
+  } catch (e) {
+    log('Ping failed: ' + e.message, 'err');
+  }
+}
+
 // --- Exec ---
 
 async function execCmd() {

internal/api/openapi.yaml

@@ -249,6 +249,40 @@ paths:
               schema:
                 $ref: "#/components/schemas/Error"
 
+  /v1/sandboxes/{id}/ping:
+    parameters:
+      - name: id
+        in: path
+        required: true
+        schema:
+          type: string
+
+    post:
+      summary: Reset sandbox inactivity timer
+      operationId: pingSandbox
+      tags: [sandboxes]
+      security:
+        - apiKeyAuth: []
+      description: |
+        Resets the last_active_at timestamp for a running sandbox, preventing
+        the auto-pause TTL from expiring. Use this as a keepalive for sandboxes
+        that are idle but should remain running.
+      responses:
+        "204":
+          description: Ping acknowledged, inactivity timer reset
+        "404":
+          description: Sandbox not found
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+        "409":
+          description: Sandbox not running
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/Error"
+
   /v1/sandboxes/{id}/pause:
     parameters:
       - name: id
@@ -721,7 +755,11 @@ components:
           default: 512
         timeout_sec:
           type: integer
-          default: 300
+          default: 0
+          description: >
+            Auto-pause TTL in seconds. The sandbox is automatically paused
+            after this duration of inactivity (no exec or ping). 0 means
+            no auto-pause.
 
     Sandbox:
       type: object

internal/api/reconciler.go

@@ -49,7 +49,8 @@ func (rc *Reconciler) Start(ctx context.Context) {
 }
 
 func (rc *Reconciler) reconcile(ctx context.Context) {
-	// Get all sandboxes the host agent knows about.
+	// Single RPC returns both the running sandbox list and any IDs that
+	// were auto-paused by the TTL reaper since the last call.
 	resp, err := rc.agent.ListSandboxes(ctx, connect.NewRequest(&pb.ListSandboxesRequest{}))
 	if err != nil {
 		slog.Warn("reconciler: failed to list sandboxes from host agent", "error", err)
@@ -62,6 +63,12 @@ func (rc *Reconciler) reconcile(ctx context.Context) {
 		alive[sb.SandboxId] = struct{}{}
 	}
 
+	// Build auto-paused set from the same response.
+	autoPausedSet := make(map[string]struct{}, len(resp.Msg.AutoPausedSandboxIds))
+	for _, id := range resp.Msg.AutoPausedSandboxIds {
+		autoPausedSet[id] = struct{}{}
+	}
+
 	// Get all DB sandboxes for this host that are running.
 	// Paused sandboxes are excluded: they are expected to not exist on the
 	// host agent because pause = snapshot + destroy resources.
@@ -86,12 +93,34 @@ func (rc *Reconciler) reconcile(ctx context.Context) {
 		return
 	}
 
-	slog.Info("reconciler: marking stale sandboxes as stopped", "count", len(stale), "ids", stale)
+	// Split stale sandboxes into those auto-paused by the TTL reaper vs
+	// those that crashed/were orphaned.
+	var toPause, toStop []string
+	for _, id := range stale {
+		if _, ok := autoPausedSet[id]; ok {
+			toPause = append(toPause, id)
+		} else {
+			toStop = append(toStop, id)
+		}
+	}
 
-	if err := rc.db.BulkUpdateStatusByIDs(ctx, db.BulkUpdateStatusByIDsParams{
-		Column1: stale,
-		Status:  "stopped",
-	}); err != nil {
-		slog.Warn("reconciler: failed to update stale sandboxes", "error", err)
+	if len(toPause) > 0 {
+		slog.Info("reconciler: marking auto-paused sandboxes", "count", len(toPause), "ids", toPause)
+		if err := rc.db.BulkUpdateStatusByIDs(ctx, db.BulkUpdateStatusByIDsParams{
+			Column1: toPause,
+			Status:  "paused",
+		}); err != nil {
+			slog.Warn("reconciler: failed to mark auto-paused sandboxes", "error", err)
+		}
+	}
+
+	if len(toStop) > 0 {
+		slog.Info("reconciler: marking stale sandboxes as stopped", "count", len(toStop), "ids", toStop)
+		if err := rc.db.BulkUpdateStatusByIDs(ctx, db.BulkUpdateStatusByIDsParams{
+			Column1: toStop,
+			Status:  "stopped",
+		}); err != nil {
+			slog.Warn("reconciler: failed to update stale sandboxes", "error", err)
+		}
 	}
 }

internal/api/server.go

@@ -64,6 +64,7 @@ func New(queries *db.Queries, agent hostagentv1connect.HostAgentServiceClient, p
 			r.Delete("/", sandbox.Destroy)
 			r.Post("/exec", exec.Exec)
 			r.Get("/exec/stream", execStream.ExecStream)
+			r.Post("/ping", sandbox.Ping)
 			r.Post("/pause", sandbox.Pause)
 			r.Post("/resume", sandbox.Resume)
 			r.Post("/files/write", files.Upload)

internal/hostagent/server.go

@@ -8,6 +8,7 @@ import (
 	"mime/multipart"
 	"net/http"
 	"net/url"
+	"strings"
 	"time"
 
 	"connectrpc.com/connect"
@@ -71,7 +72,7 @@ func (s *Server) ResumeSandbox(
 	ctx context.Context,
 	req *connect.Request[pb.ResumeSandboxRequest],
 ) (*connect.Response[pb.ResumeSandboxResponse], error) {
-	sb, err := s.mgr.Resume(ctx, req.Msg.SandboxId)
+	sb, err := s.mgr.Resume(ctx, req.Msg.SandboxId, int(req.Msg.TimeoutSec))
 	if err != nil {
 		return nil, connect.NewError(connect.CodeInternal, err)
 	}
@@ -106,6 +107,19 @@ func (s *Server) DeleteSnapshot(
 	return connect.NewResponse(&pb.DeleteSnapshotResponse{}), nil
 }
 
+func (s *Server) PingSandbox(
+	ctx context.Context,
+	req *connect.Request[pb.PingSandboxRequest],
+) (*connect.Response[pb.PingSandboxResponse], error) {
+	if err := s.mgr.Ping(req.Msg.SandboxId); err != nil {
+		if strings.Contains(err.Error(), "not found") {
+			return nil, connect.NewError(connect.CodeNotFound, err)
+		}
+		return nil, connect.NewError(connect.CodeFailedPrecondition, err)
+	}
+	return connect.NewResponse(&pb.PingSandboxResponse{}), nil
+}
+
 func (s *Server) Exec(
 	ctx context.Context,
 	req *connect.Request[pb.ExecRequest],
@@ -394,6 +408,7 @@ func (s *Server) ListSandboxes(
 	}
 
 	return connect.NewResponse(&pb.ListSandboxesResponse{
-		Sandboxes: infos,
+		Sandboxes:            infos,
+		AutoPausedSandboxIds: s.mgr.DrainAutoPausedIDs(),
 	}), nil
 }

internal/sandbox/manager.go

@@ -40,6 +40,9 @@ type Manager struct {
 	mu     sync.RWMutex
 	boxes  map[string]*sandboxState
 	stopCh chan struct{}
+
+	autoPausedMu  sync.Mutex
+	autoPausedIDs []string
 }
 
 // sandboxState holds the runtime state for a single sandbox.
@@ -459,7 +462,7 @@ func (m *Manager) Pause(ctx context.Context, sandboxID string) error {
 
 // Resume restores a paused sandbox from its snapshot using UFFD for
 // lazy memory loading. The sandbox gets a new network slot.
-func (m *Manager) Resume(ctx context.Context, sandboxID string) (*models.Sandbox, error) {
+func (m *Manager) Resume(ctx context.Context, sandboxID string, timeoutSec int) (*models.Sandbox, error) {
 	snapDir := m.cfg.SnapshotsDir
 	if !snapshot.Exists(snapDir, sandboxID) {
 		return nil, fmt.Errorf("no snapshot found for sandbox %s", sandboxID)
@@ -575,7 +578,7 @@ func (m *Manager) Resume(ctx context.Context, sandboxID string) (*models.Sandbox
 		SandboxID:        sandboxID,
 		KernelPath:       m.cfg.KernelPath,
 		RootfsPath:       dmDev.DevicePath,
-		VCPUs:            1,                                          // Placeholder; overridden by snapshot.
+		VCPUs:            1,                                         // Placeholder; overridden by snapshot.
 		MemoryMB:         int(header.Metadata.Size / (1024 * 1024)), // Placeholder; overridden by snapshot.
 		NetworkNamespace: slot.NamespaceID,
 		TapDevice:        slot.TapName,
@@ -622,7 +625,7 @@ func (m *Manager) Resume(ctx context.Context, sandboxID string) (*models.Sandbox
 			Template:     "",
 			VCPUs:        vmCfg.VCPUs,
 			MemoryMB:     vmCfg.MemoryMB,
-			TimeoutSec:   0,
+			TimeoutSec:   timeoutSec,
 			SlotIndex:    slotIdx,
 			HostIP:       slot.HostIP,
 			RootfsPath:   dmDev.DevicePath,
@@ -1033,6 +1036,33 @@ func (m *Manager) GetClient(sandboxID string) (*envdclient.Client, error) {
 	return sb.client, nil
 }
 
+// Ping resets the inactivity timer for a running sandbox.
+func (m *Manager) Ping(sandboxID string) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	sb, ok := m.boxes[sandboxID]
+	if !ok {
+		return fmt.Errorf("sandbox not found: %s", sandboxID)
+	}
+	if sb.Status != models.StatusRunning {
+		return fmt.Errorf("sandbox %s is not running (status: %s)", sandboxID, sb.Status)
+	}
+	sb.LastActiveAt = time.Now()
+	return nil
+}
+
+// DrainAutoPausedIDs returns and clears the list of sandbox IDs that were
+// automatically paused by the TTL reaper since the last call.
+func (m *Manager) DrainAutoPausedIDs() []string {
+	m.autoPausedMu.Lock()
+	defer m.autoPausedMu.Unlock()
+
+	ids := m.autoPausedIDs
+	m.autoPausedIDs = nil
+	return ids
+}
+
 func (m *Manager) get(sandboxID string) (*sandboxState, error) {
 	m.mu.RLock()
 	defer m.mu.RUnlock()
@@ -1048,7 +1078,7 @@ func (m *Manager) get(sandboxID string) (*sandboxState, error) {
 // that have exceeded their TTL (timeout_sec of inactivity).
 func (m *Manager) StartTTLReaper(ctx context.Context) {
 	go func() {
-		ticker := time.NewTicker(10 * time.Second)
+		ticker := time.NewTicker(2 * time.Second)
 		defer ticker.Stop()
 
 		for {
@@ -1064,7 +1094,7 @@ func (m *Manager) StartTTLReaper(ctx context.Context) {
 	}()
 }
 
-func (m *Manager) reapExpired(ctx context.Context) {
+func (m *Manager) reapExpired(_ context.Context) {
 	m.mu.RLock()
 	var expired []string
 	now := time.Now()
@@ -1072,7 +1102,7 @@ func (m *Manager) reapExpired(ctx context.Context) {
 		if sb.TimeoutSec <= 0 {
 			continue
 		}
-		if sb.Status != models.StatusRunning && sb.Status != models.StatusPaused {
+		if sb.Status != models.StatusRunning {
 			continue
 		}
 		if now.Sub(sb.LastActiveAt) > time.Duration(sb.TimeoutSec)*time.Second {
@@ -1082,10 +1112,23 @@ func (m *Manager) reapExpired(ctx context.Context) {
 	m.mu.RUnlock()
 
 	for _, id := range expired {
-		slog.Info("TTL expired, destroying sandbox", "id", id)
-		if err := m.Destroy(ctx, id); err != nil {
-			slog.Warn("TTL reap failed", "id", id, "error", err)
+		slog.Info("TTL expired, auto-pausing sandbox", "id", id)
+		// Use a detached context so that an app shutdown does not cancel
+		// a pause mid-flight, which would leave the VM frozen without a
+		// valid snapshot.
+		pauseCtx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
+		err := m.Pause(pauseCtx, id)
+		cancel()
+		if err != nil {
+			slog.Warn("TTL auto-pause failed, destroying sandbox", "id", id, "error", err)
+			if destroyErr := m.Destroy(context.Background(), id); destroyErr != nil {
+				slog.Warn("TTL destroy after failed pause also failed", "id", id, "error", destroyErr)
+			}
+			continue
 		}
+		m.autoPausedMu.Lock()
+		m.autoPausedIDs = append(m.autoPausedIDs, id)
+		m.autoPausedMu.Unlock()
 	}
 }
 

proto/hostagent/gen/hostagent.pb.go

@@ -31,8 +31,8 @@ type CreateSandboxRequest struct {
 	Vcpus int32 `protobuf:"varint,2,opt,name=vcpus,proto3" json:"vcpus,omitempty"`
 	// Memory in MB (default: 512).
 	MemoryMb int32 `protobuf:"varint,3,opt,name=memory_mb,json=memoryMb,proto3" json:"memory_mb,omitempty"`
-	// TTL in seconds. Sandbox is auto-destroyed after this duration of
-	// inactivity. 0 means no auto-destroy.
+	// TTL in seconds. Sandbox is auto-paused after this duration of
+	// inactivity. 0 means no auto-pause.
 	TimeoutSec    int32 `protobuf:"varint,4,opt,name=timeout_sec,json=timeoutSec,proto3" json:"timeout_sec,omitempty"`
 	unknownFields protoimpl.UnknownFields
 	sizeCache     protoimpl.SizeCache
@@ -324,8 +324,11 @@ func (*PauseSandboxResponse) Descriptor() ([]byte, []int) {
 }
 
 type ResumeSandboxRequest struct {
-	state         protoimpl.MessageState `protogen:"open.v1"`
-	SandboxId     string                 `protobuf:"bytes,1,opt,name=sandbox_id,json=sandboxId,proto3" json:"sandbox_id,omitempty"`
+	state     protoimpl.MessageState `protogen:"open.v1"`
+	SandboxId string                 `protobuf:"bytes,1,opt,name=sandbox_id,json=sandboxId,proto3" json:"sandbox_id,omitempty"`
+	// TTL in seconds restored from the DB so the reaper can auto-pause
+	// the sandbox again after inactivity. 0 means no auto-pause.
+	TimeoutSec    int32 `protobuf:"varint,2,opt,name=timeout_sec,json=timeoutSec,proto3" json:"timeout_sec,omitempty"`
 	unknownFields protoimpl.UnknownFields
 	sizeCache     protoimpl.SizeCache
 }
@@ -367,6 +370,13 @@ func (x *ResumeSandboxRequest) GetSandboxId() string {
 	return ""
 }
 
+func (x *ResumeSandboxRequest) GetTimeoutSec() int32 {
+	if x != nil {
+		return x.TimeoutSec
+	}
+	return 0
+}
+
 type ResumeSandboxResponse struct {
 	state         protoimpl.MessageState `protogen:"open.v1"`
 	SandboxId     string                 `protobuf:"bytes,1,opt,name=sandbox_id,json=sandboxId,proto3" json:"sandbox_id,omitempty"`
@@ -777,10 +787,13 @@ func (*ListSandboxesRequest) Descriptor() ([]byte, []int) {
 }
 
 type ListSandboxesResponse struct {
-	state         protoimpl.MessageState `protogen:"open.v1"`
-	Sandboxes     []*SandboxInfo         `protobuf:"bytes,1,rep,name=sandboxes,proto3" json:"sandboxes,omitempty"`
-	unknownFields protoimpl.UnknownFields
-	sizeCache     protoimpl.SizeCache
+	state     protoimpl.MessageState `protogen:"open.v1"`
+	Sandboxes []*SandboxInfo         `protobuf:"bytes,1,rep,name=sandboxes,proto3" json:"sandboxes,omitempty"`
+	// IDs of sandboxes that were automatically paused by the TTL reaper
+	// since the last call. Drained on read.
+	AutoPausedSandboxIds []string `protobuf:"bytes,2,rep,name=auto_paused_sandbox_ids,json=autoPausedSandboxIds,proto3" json:"auto_paused_sandbox_ids,omitempty"`
+	unknownFields        protoimpl.UnknownFields
+	sizeCache            protoimpl.SizeCache
 }
 
 func (x *ListSandboxesResponse) Reset() {
@@ -820,6 +833,13 @@ func (x *ListSandboxesResponse) GetSandboxes() []*SandboxInfo {
 	return nil
 }
 
+func (x *ListSandboxesResponse) GetAutoPausedSandboxIds() []string {
+	if x != nil {
+		return x.AutoPausedSandboxIds
+	}
+	return nil
+}
+
 type SandboxInfo struct {
 	state            protoimpl.MessageState `protogen:"open.v1"`
 	SandboxId        string                 `protobuf:"bytes,1,opt,name=sandbox_id,json=sandboxId,proto3" json:"sandbox_id,omitempty"`
@@ -1730,6 +1750,86 @@ func (x *ReadFileStreamResponse) GetChunk() []byte {
 	return nil
 }
 
+type PingSandboxRequest struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	SandboxId     string                 `protobuf:"bytes,1,opt,name=sandbox_id,json=sandboxId,proto3" json:"sandbox_id,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *PingSandboxRequest) Reset() {
+	*x = PingSandboxRequest{}
+	mi := &file_hostagent_proto_msgTypes[31]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *PingSandboxRequest) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*PingSandboxRequest) ProtoMessage() {}
+
+func (x *PingSandboxRequest) ProtoReflect() protoreflect.Message {
+	mi := &file_hostagent_proto_msgTypes[31]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use PingSandboxRequest.ProtoReflect.Descriptor instead.
+func (*PingSandboxRequest) Descriptor() ([]byte, []int) {
+	return file_hostagent_proto_rawDescGZIP(), []int{31}
+}
+
+func (x *PingSandboxRequest) GetSandboxId() string {
+	if x != nil {
+		return x.SandboxId
+	}
+	return ""
+}
+
+type PingSandboxResponse struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *PingSandboxResponse) Reset() {
+	*x = PingSandboxResponse{}
+	mi := &file_hostagent_proto_msgTypes[32]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *PingSandboxResponse) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*PingSandboxResponse) ProtoMessage() {}
+
+func (x *PingSandboxResponse) ProtoReflect() protoreflect.Message {
+	mi := &file_hostagent_proto_msgTypes[32]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use PingSandboxResponse.ProtoReflect.Descriptor instead.
+func (*PingSandboxResponse) Descriptor() ([]byte, []int) {
+	return file_hostagent_proto_rawDescGZIP(), []int{32}
+}
+
 var File_hostagent_proto protoreflect.FileDescriptor
 
 const file_hostagent_proto_rawDesc = "" +
@@ -1755,10 +1855,12 @@ const file_hostagent_proto_rawDesc = "" +
 	"\x13PauseSandboxRequest\x12\x1d\n" +
 	"\n" +
 	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\"\x16\n" +
-	"\x14PauseSandboxResponse\"5\n" +
+	"\x14PauseSandboxResponse\"V\n" +
 	"\x14ResumeSandboxRequest\x12\x1d\n" +
 	"\n" +
-	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\"g\n" +
+	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\x12\x1f\n" +
+	"\vtimeout_sec\x18\x02 \x01(\x05R\n" +
+	"timeoutSec\"g\n" +
 	"\x15ResumeSandboxResponse\x12\x1d\n" +
 	"\n" +
 	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\x12\x16\n" +
@@ -1786,9 +1888,10 @@ const file_hostagent_proto_rawDesc = "" +
 	"\x06stdout\x18\x01 \x01(\fR\x06stdout\x12\x16\n" +
 	"\x06stderr\x18\x02 \x01(\fR\x06stderr\x12\x1b\n" +
 	"\texit_code\x18\x03 \x01(\x05R\bexitCode\"\x16\n" +
-	"\x14ListSandboxesRequest\"P\n" +
+	"\x14ListSandboxesRequest\"\x87\x01\n" +
 	"\x15ListSandboxesResponse\x127\n" +
-	"\tsandboxes\x18\x01 \x03(\v2\x19.hostagent.v1.SandboxInfoR\tsandboxes\"\xa4\x02\n" +
+	"\tsandboxes\x18\x01 \x03(\v2\x19.hostagent.v1.SandboxInfoR\tsandboxes\x125\n" +
+	"\x17auto_paused_sandbox_ids\x18\x02 \x03(\tR\x14autoPausedSandboxIds\"\xa4\x02\n" +
 	"\vSandboxInfo\x12\x1d\n" +
 	"\n" +
 	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\x12\x16\n" +
@@ -1848,7 +1951,11 @@ const file_hostagent_proto_rawDesc = "" +
 	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\x12\x12\n" +
 	"\x04path\x18\x02 \x01(\tR\x04path\".\n" +
 	"\x16ReadFileStreamResponse\x12\x14\n" +
-	"\x05chunk\x18\x01 \x01(\fR\x05chunk2\xfa\b\n" +
+	"\x05chunk\x18\x01 \x01(\fR\x05chunk\"3\n" +
+	"\x12PingSandboxRequest\x12\x1d\n" +
+	"\n" +
+	"sandbox_id\x18\x01 \x01(\tR\tsandboxId\"\x15\n" +
+	"\x13PingSandboxResponse2\xce\t\n" +
 	"\x10HostAgentService\x12X\n" +
 	"\rCreateSandbox\x12\".hostagent.v1.CreateSandboxRequest\x1a#.hostagent.v1.CreateSandboxResponse\x12[\n" +
 	"\x0eDestroySandbox\x12#.hostagent.v1.DestroySandboxRequest\x1a$.hostagent.v1.DestroySandboxResponse\x12U\n" +
@@ -1863,7 +1970,8 @@ const file_hostagent_proto_rawDesc = "" +
 	"\n" +
 	"ExecStream\x12\x1f.hostagent.v1.ExecStreamRequest\x1a .hostagent.v1.ExecStreamResponse0\x01\x12`\n" +
 	"\x0fWriteFileStream\x12$.hostagent.v1.WriteFileStreamRequest\x1a%.hostagent.v1.WriteFileStreamResponse(\x01\x12]\n" +
-	"\x0eReadFileStream\x12#.hostagent.v1.ReadFileStreamRequest\x1a$.hostagent.v1.ReadFileStreamResponse0\x01B\xb0\x01\n" +
+	"\x0eReadFileStream\x12#.hostagent.v1.ReadFileStreamRequest\x1a$.hostagent.v1.ReadFileStreamResponse0\x01\x12R\n" +
+	"\vPingSandbox\x12 .hostagent.v1.PingSandboxRequest\x1a!.hostagent.v1.PingSandboxResponseB\xb0\x01\n" +
 	"\x10com.hostagent.v1B\x0eHostagentProtoP\x01Z;git.omukk.dev/wrenn/sandbox/proto/hostagent/gen;hostagentv1\xa2\x02\x03HXX\xaa\x02\fHostagent.V1\xca\x02\fHostagent\\V1\xe2\x02\x18Hostagent\\V1\\GPBMetadata\xea\x02\rHostagent::V1b\x06proto3"
 
 var (
@@ -1878,7 +1986,7 @@ func file_hostagent_proto_rawDescGZIP() []byte {
 	return file_hostagent_proto_rawDescData
 }
 
-var file_hostagent_proto_msgTypes = make([]protoimpl.MessageInfo, 31)
+var file_hostagent_proto_msgTypes = make([]protoimpl.MessageInfo, 33)
 var file_hostagent_proto_goTypes = []any{
 	(*CreateSandboxRequest)(nil),    // 0: hostagent.v1.CreateSandboxRequest
 	(*CreateSandboxResponse)(nil),   // 1: hostagent.v1.CreateSandboxResponse
@@ -1911,6 +2019,8 @@ var file_hostagent_proto_goTypes = []any{
 	(*WriteFileStreamResponse)(nil), // 28: hostagent.v1.WriteFileStreamResponse
 	(*ReadFileStreamRequest)(nil),   // 29: hostagent.v1.ReadFileStreamRequest
 	(*ReadFileStreamResponse)(nil),  // 30: hostagent.v1.ReadFileStreamResponse
+	(*PingSandboxRequest)(nil),      // 31: hostagent.v1.PingSandboxRequest
+	(*PingSandboxResponse)(nil),     // 32: hostagent.v1.PingSandboxResponse
 }
 var file_hostagent_proto_depIdxs = []int32{
 	16, // 0: hostagent.v1.ListSandboxesResponse.sandboxes:type_name -> hostagent.v1.SandboxInfo
@@ -1931,21 +2041,23 @@ var file_hostagent_proto_depIdxs = []int32{
 	21, // 15: hostagent.v1.HostAgentService.ExecStream:input_type -> hostagent.v1.ExecStreamRequest
 	26, // 16: hostagent.v1.HostAgentService.WriteFileStream:input_type -> hostagent.v1.WriteFileStreamRequest
 	29, // 17: hostagent.v1.HostAgentService.ReadFileStream:input_type -> hostagent.v1.ReadFileStreamRequest
-	1,  // 18: hostagent.v1.HostAgentService.CreateSandbox:output_type -> hostagent.v1.CreateSandboxResponse
-	3,  // 19: hostagent.v1.HostAgentService.DestroySandbox:output_type -> hostagent.v1.DestroySandboxResponse
-	5,  // 20: hostagent.v1.HostAgentService.PauseSandbox:output_type -> hostagent.v1.PauseSandboxResponse
-	7,  // 21: hostagent.v1.HostAgentService.ResumeSandbox:output_type -> hostagent.v1.ResumeSandboxResponse
-	13, // 22: hostagent.v1.HostAgentService.Exec:output_type -> hostagent.v1.ExecResponse
-	15, // 23: hostagent.v1.HostAgentService.ListSandboxes:output_type -> hostagent.v1.ListSandboxesResponse
-	18, // 24: hostagent.v1.HostAgentService.WriteFile:output_type -> hostagent.v1.WriteFileResponse
-	20, // 25: hostagent.v1.HostAgentService.ReadFile:output_type -> hostagent.v1.ReadFileResponse
-	9,  // 26: hostagent.v1.HostAgentService.CreateSnapshot:output_type -> hostagent.v1.CreateSnapshotResponse
-	11, // 27: hostagent.v1.HostAgentService.DeleteSnapshot:output_type -> hostagent.v1.DeleteSnapshotResponse
-	22, // 28: hostagent.v1.HostAgentService.ExecStream:output_type -> hostagent.v1.ExecStreamResponse
-	28, // 29: hostagent.v1.HostAgentService.WriteFileStream:output_type -> hostagent.v1.WriteFileStreamResponse
-	30, // 30: hostagent.v1.HostAgentService.ReadFileStream:output_type -> hostagent.v1.ReadFileStreamResponse
-	18, // [18:31] is the sub-list for method output_type
-	5,  // [5:18] is the sub-list for method input_type
+	31, // 18: hostagent.v1.HostAgentService.PingSandbox:input_type -> hostagent.v1.PingSandboxRequest
+	1,  // 19: hostagent.v1.HostAgentService.CreateSandbox:output_type -> hostagent.v1.CreateSandboxResponse
+	3,  // 20: hostagent.v1.HostAgentService.DestroySandbox:output_type -> hostagent.v1.DestroySandboxResponse
+	5,  // 21: hostagent.v1.HostAgentService.PauseSandbox:output_type -> hostagent.v1.PauseSandboxResponse
+	7,  // 22: hostagent.v1.HostAgentService.ResumeSandbox:output_type -> hostagent.v1.ResumeSandboxResponse
+	13, // 23: hostagent.v1.HostAgentService.Exec:output_type -> hostagent.v1.ExecResponse
+	15, // 24: hostagent.v1.HostAgentService.ListSandboxes:output_type -> hostagent.v1.ListSandboxesResponse
+	18, // 25: hostagent.v1.HostAgentService.WriteFile:output_type -> hostagent.v1.WriteFileResponse
+	20, // 26: hostagent.v1.HostAgentService.ReadFile:output_type -> hostagent.v1.ReadFileResponse
+	9,  // 27: hostagent.v1.HostAgentService.CreateSnapshot:output_type -> hostagent.v1.CreateSnapshotResponse
+	11, // 28: hostagent.v1.HostAgentService.DeleteSnapshot:output_type -> hostagent.v1.DeleteSnapshotResponse
+	22, // 29: hostagent.v1.HostAgentService.ExecStream:output_type -> hostagent.v1.ExecStreamResponse
+	28, // 30: hostagent.v1.HostAgentService.WriteFileStream:output_type -> hostagent.v1.WriteFileStreamResponse
+	30, // 31: hostagent.v1.HostAgentService.ReadFileStream:output_type -> hostagent.v1.ReadFileStreamResponse
+	32, // 32: hostagent.v1.HostAgentService.PingSandbox:output_type -> hostagent.v1.PingSandboxResponse
+	19, // [19:33] is the sub-list for method output_type
+	5,  // [5:19] is the sub-list for method input_type
 	5,  // [5:5] is the sub-list for extension type_name
 	5,  // [5:5] is the sub-list for extension extendee
 	0,  // [0:5] is the sub-list for field type_name
@@ -1975,7 +2087,7 @@ func file_hostagent_proto_init() {
 			GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
 			RawDescriptor: unsafe.Slice(unsafe.StringData(file_hostagent_proto_rawDesc), len(file_hostagent_proto_rawDesc)),
 			NumEnums:      0,
-			NumMessages:   31,
+			NumMessages:   33,
 			NumExtensions: 0,
 			NumServices:   1,
 		},

proto/hostagent/gen/hostagentv1connect/hostagent.connect.go

@@ -71,6 +71,9 @@ const (
 	// HostAgentServiceReadFileStreamProcedure is the fully-qualified name of the HostAgentService's
 	// ReadFileStream RPC.
 	HostAgentServiceReadFileStreamProcedure = "/hostagent.v1.HostAgentService/ReadFileStream"
+	// HostAgentServicePingSandboxProcedure is the fully-qualified name of the HostAgentService's
+	// PingSandbox RPC.
+	HostAgentServicePingSandboxProcedure = "/hostagent.v1.HostAgentService/PingSandbox"
 )
 
 // HostAgentServiceClient is a client for the hostagent.v1.HostAgentService service.
@@ -103,6 +106,8 @@ type HostAgentServiceClient interface {
 	WriteFileStream(context.Context) *connect.ClientStreamForClient[gen.WriteFileStreamRequest, gen.WriteFileStreamResponse]
 	// ReadFileStream reads a file from a sandbox and streams it back in chunks.
 	ReadFileStream(context.Context, *connect.Request[gen.ReadFileStreamRequest]) (*connect.ServerStreamForClient[gen.ReadFileStreamResponse], error)
+	// PingSandbox resets the inactivity timer for a running sandbox.
+	PingSandbox(context.Context, *connect.Request[gen.PingSandboxRequest]) (*connect.Response[gen.PingSandboxResponse], error)
 }
 
 // NewHostAgentServiceClient constructs a client for the hostagent.v1.HostAgentService service. By
@@ -194,6 +199,12 @@ func NewHostAgentServiceClient(httpClient connect.HTTPClient, baseURL string, op
 			connect.WithSchema(hostAgentServiceMethods.ByName("ReadFileStream")),
 			connect.WithClientOptions(opts...),
 		),
+		pingSandbox: connect.NewClient[gen.PingSandboxRequest, gen.PingSandboxResponse](
+			httpClient,
+			baseURL+HostAgentServicePingSandboxProcedure,
+			connect.WithSchema(hostAgentServiceMethods.ByName("PingSandbox")),
+			connect.WithClientOptions(opts...),
+		),
 	}
 }
 
@@ -212,6 +223,7 @@ type hostAgentServiceClient struct {
 	execStream      *connect.Client[gen.ExecStreamRequest, gen.ExecStreamResponse]
 	writeFileStream *connect.Client[gen.WriteFileStreamRequest, gen.WriteFileStreamResponse]
 	readFileStream  *connect.Client[gen.ReadFileStreamRequest, gen.ReadFileStreamResponse]
+	pingSandbox     *connect.Client[gen.PingSandboxRequest, gen.PingSandboxResponse]
 }
 
 // CreateSandbox calls hostagent.v1.HostAgentService.CreateSandbox.
@@ -279,6 +291,11 @@ func (c *hostAgentServiceClient) ReadFileStream(ctx context.Context, req *connec
 	return c.readFileStream.CallServerStream(ctx, req)
 }
 
+// PingSandbox calls hostagent.v1.HostAgentService.PingSandbox.
+func (c *hostAgentServiceClient) PingSandbox(ctx context.Context, req *connect.Request[gen.PingSandboxRequest]) (*connect.Response[gen.PingSandboxResponse], error) {
+	return c.pingSandbox.CallUnary(ctx, req)
+}
+
 // HostAgentServiceHandler is an implementation of the hostagent.v1.HostAgentService service.
 type HostAgentServiceHandler interface {
 	// CreateSandbox boots a new microVM with the given configuration.
@@ -309,6 +326,8 @@ type HostAgentServiceHandler interface {
 	WriteFileStream(context.Context, *connect.ClientStream[gen.WriteFileStreamRequest]) (*connect.Response[gen.WriteFileStreamResponse], error)
 	// ReadFileStream reads a file from a sandbox and streams it back in chunks.
 	ReadFileStream(context.Context, *connect.Request[gen.ReadFileStreamRequest], *connect.ServerStream[gen.ReadFileStreamResponse]) error
+	// PingSandbox resets the inactivity timer for a running sandbox.
+	PingSandbox(context.Context, *connect.Request[gen.PingSandboxRequest]) (*connect.Response[gen.PingSandboxResponse], error)
 }
 
 // NewHostAgentServiceHandler builds an HTTP handler from the service implementation. It returns the
@@ -396,6 +415,12 @@ func NewHostAgentServiceHandler(svc HostAgentServiceHandler, opts ...connect.Han
 		connect.WithSchema(hostAgentServiceMethods.ByName("ReadFileStream")),
 		connect.WithHandlerOptions(opts...),
 	)
+	hostAgentServicePingSandboxHandler := connect.NewUnaryHandler(
+		HostAgentServicePingSandboxProcedure,
+		svc.PingSandbox,
+		connect.WithSchema(hostAgentServiceMethods.ByName("PingSandbox")),
+		connect.WithHandlerOptions(opts...),
+	)
 	return "/hostagent.v1.HostAgentService/", http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		switch r.URL.Path {
 		case HostAgentServiceCreateSandboxProcedure:
@@ -424,6 +449,8 @@ func NewHostAgentServiceHandler(svc HostAgentServiceHandler, opts ...connect.Han
 			hostAgentServiceWriteFileStreamHandler.ServeHTTP(w, r)
 		case HostAgentServiceReadFileStreamProcedure:
 			hostAgentServiceReadFileStreamHandler.ServeHTTP(w, r)
+		case HostAgentServicePingSandboxProcedure:
+			hostAgentServicePingSandboxHandler.ServeHTTP(w, r)
 		default:
 			http.NotFound(w, r)
 		}
@@ -484,3 +511,7 @@ func (UnimplementedHostAgentServiceHandler) WriteFileStream(context.Context, *co
 func (UnimplementedHostAgentServiceHandler) ReadFileStream(context.Context, *connect.Request[gen.ReadFileStreamRequest], *connect.ServerStream[gen.ReadFileStreamResponse]) error {
 	return connect.NewError(connect.CodeUnimplemented, errors.New("hostagent.v1.HostAgentService.ReadFileStream is not implemented"))
 }
+
+func (UnimplementedHostAgentServiceHandler) PingSandbox(context.Context, *connect.Request[gen.PingSandboxRequest]) (*connect.Response[gen.PingSandboxResponse], error) {
+	return nil, connect.NewError(connect.CodeUnimplemented, errors.New("hostagent.v1.HostAgentService.PingSandbox is not implemented"))
+}

proto/hostagent/hostagent.proto

@@ -45,6 +45,10 @@ service HostAgentService {
 
   // ReadFileStream reads a file from a sandbox and streams it back in chunks.
   rpc ReadFileStream(ReadFileStreamRequest) returns (stream ReadFileStreamResponse);
+
+  // PingSandbox resets the inactivity timer for a running sandbox.
+  rpc PingSandbox(PingSandboxRequest) returns (PingSandboxResponse);
+
 }
 
 message CreateSandboxRequest {
@@ -60,8 +64,8 @@ message CreateSandboxRequest {
   // Memory in MB (default: 512).
   int32 memory_mb = 3;
 
-  // TTL in seconds. Sandbox is auto-destroyed after this duration of
-  // inactivity. 0 means no auto-destroy.
+  // TTL in seconds. Sandbox is auto-paused after this duration of
+  // inactivity. 0 means no auto-pause.
   int32 timeout_sec = 4;
 }
 
@@ -85,6 +89,10 @@ message PauseSandboxResponse {}
 
 message ResumeSandboxRequest {
   string sandbox_id = 1;
+
+  // TTL in seconds restored from the DB so the reaper can auto-pause
+  // the sandbox again after inactivity. 0 means no auto-pause.
+  int32 timeout_sec = 2;
 }
 
 message ResumeSandboxResponse {
@@ -127,6 +135,10 @@ message ListSandboxesRequest {}
 
 message ListSandboxesResponse {
   repeated SandboxInfo sandboxes = 1;
+
+  // IDs of sandboxes that were automatically paused by the TTL reaper
+  // since the last call. Drained on read.
+  repeated string auto_paused_sandbox_ids = 2;
 }
 
 message SandboxInfo {
@@ -215,3 +227,12 @@ message ReadFileStreamRequest {
 message ReadFileStreamResponse {
   bytes chunk = 1;
 }
+
+// ── Ping ────────────────────────────────────────────────────────────
+
+message PingSandboxRequest {
+  string sandbox_id = 1;
+}
+
+message PingSandboxResponse {}
+