feat(runtime): track node lifecycle and QUIC link state

backend/shared/MoShiro.Protocol/Rpc/MoShiroMethodTable.cs

@@ -30,6 +30,7 @@ public static class MoShiroMethodTable
                 MethodSpec.Request(MoShiroMethods.TestRun, RpcSide.Server, NodeRoles.Agent),
                 MethodSpec.Request(MoShiroMethods.TestCancel, RpcSide.Server, NodeRoles.Agent),
                 MethodSpec.Request(MoShiroMethods.ProbeDisconnect, RpcSide.Server, NodeRoles.Agent),
+                MethodSpec.Notification(MoShiroMethods.ProbeLinkRetry, RpcSide.Server, NodeRoles.Probe),
                 MethodSpec.Notification(MoShiroMethods.UpdateOffer, RpcSide.Server),
                 MethodSpec.Notification(MoShiroMethods.UpdateApply, RpcSide.Server),
                 MethodSpec.Notification(MoShiroMethods.Schedule, RpcSide.Server),

backend/shared/MoShiro.Protocol/Rpc/MoShiroRpc.cs

@@ -10,6 +10,7 @@ public static class MoShiroMethods
     public const string TestRun = "moshiro.test.run";
     public const string TestCancel = "moshiro.test.cancel";
     public const string ProbeDisconnect = "moshiro.probe.disconnect";
+    public const string ProbeLinkRetry = "moshiro.probe.link_retry";
     public const string TestStatus = "moshiro.test.status";
     public const string Event = "moshiro.event";
     public const string Host = "moshiro.host";
@@ -98,6 +99,13 @@ public sealed class ProbeDisconnectParams
     public ProbeId ProbeId { get; set; }
 }
 
+/// <summary>Backend → Probe: immediately retry the selected Agent's current IP paths.</summary>
+public sealed class ProbeLinkRetryParams
+{
+    [JsonPropertyName("agent_id")]
+    public NodeId AgentId { get; set; }
+}
+
 public sealed class TestStatusParams
 {
     [JsonPropertyName("test_id")]
@@ -302,6 +310,10 @@ public sealed class UpdateStatus
 /// <summary>Probe 的 <c>fleet.heartbeat</c> 附加字段：更新器的状态。</summary>
 public sealed class ProbeHeartbeatExtra
 {
+    /// <summary>Node operating state; registration approval and transport presence are reported separately.</summary>
+    [JsonPropertyName("runtime_status")]
+    public NodeRuntimeStatus RuntimeStatus { get; set; } = NodeRuntimeStatus.Running;
+
     [JsonPropertyName("quic_links")]
     public List<ProbeQuicLinkHealth>? QuicLinks { get; set; }
 
@@ -324,6 +336,18 @@ public sealed class ProbeQuicLinkHealth
     [JsonPropertyName("connected")]
     public bool Connected { get; set; }
 
+    [JsonPropertyName("state")]
+    public ProbeLinkAttemptState State { get; set; } = ProbeLinkAttemptState.Connecting;
+
+    [JsonPropertyName("last_error")]
+    public string? LastError { get; set; }
+
+    [JsonPropertyName("retry_attempt")]
+    public int RetryAttempt { get; set; }
+
+    [JsonPropertyName("retry_in_ms")]
+    public long? RetryInMs { get; set; }
+
     [JsonPropertyName("handshake_attempts")]
     public long HandshakeAttempts { get; set; }
 
@@ -359,6 +383,17 @@ public sealed class ProbeQuicLinkHealth
     public ushort? PathMtu { get; set; }
 }
 
+[JsonConverter(typeof(JsonStringEnumConverter<ProbeLinkAttemptState>))]
+public enum ProbeLinkAttemptState
+{
+    [JsonStringEnumMemberName("connecting")]
+    Connecting,
+    [JsonStringEnumMemberName("retry_wait")]
+    RetryWait,
+    [JsonStringEnumMemberName("connected")]
+    Connected,
+}
+
 /// <summary>Probe 的 <c>fleet.hello</c> 附加字段：系统、架构和构建信息（只用于展示和更新）。</summary>
 public sealed class ProbeHelloMeta
 {
@@ -479,6 +514,10 @@ public sealed class ConnectedProbe
 /// <summary><c>fleet.heartbeat</c> 的附加字段：只有进程内的计数，没有主机信息。</summary>
 public sealed class AgentHeartbeatExtra
 {
+    /// <summary>Node operating state; registration approval and transport presence are reported separately.</summary>
+    [JsonPropertyName("runtime_status")]
+    public NodeRuntimeStatus RuntimeStatus { get; set; } = NodeRuntimeStatus.Running;
+
     [JsonPropertyName("ipv6")]
     public Ipv6Diagnostic? Ipv6 { get; set; }
 
@@ -645,6 +684,7 @@ public static class FragmentKeys
 [JsonSerializable(typeof(TestRunResult))]
 [JsonSerializable(typeof(TestCancelParams))]
 [JsonSerializable(typeof(ProbeDisconnectParams))]
+[JsonSerializable(typeof(ProbeLinkRetryParams))]
 [JsonSerializable(typeof(HostSnapshot))]
 [JsonSerializable(typeof(TestStatusParams))]
 [JsonSerializable(typeof(AgentEventParams))]

backend/shared/MoShiro.Protocol/Rpc/NodeRuntimeStatus.cs

@@ -0,0 +1,23 @@
+using System.Text.Json.Serialization;
+
+namespace MoShiro.Protocol;
+
+/// <summary>Live process/measurement state. Enrollment approval remains a separate node status.</summary>
+[JsonConverter(typeof(JsonStringEnumConverter<NodeRuntimeStatus>))]
+public enum NodeRuntimeStatus
+{
+    [JsonStringEnumMemberName("initializing")]
+    Initializing,
+    [JsonStringEnumMemberName("running")]
+    Running,
+    [JsonStringEnumMemberName("probe_paused")]
+    ProbePaused,
+    [JsonStringEnumMemberName("maintenance")]
+    Maintenance,
+    [JsonStringEnumMemberName("stopping")]
+    Stopping,
+    [JsonStringEnumMemberName("offline")]
+    Offline,
+    [JsonStringEnumMemberName("updating")]
+    Updating,
+}

backend/shared/tests/MoShiro.Protocol.Tests/NodeRuntimeStatusTests.cs

@@ -0,0 +1,17 @@
+using System.Text.Json;
+
+namespace MoShiro.Protocol.Tests;
+
+public sealed class NodeRuntimeStatusTests
+{
+    [Fact]
+    public void HeartbeatStatus_UsesStableSnakeCaseEnumValues()
+    {
+        var heartbeat = new AgentHeartbeatExtra { RuntimeStatus = NodeRuntimeStatus.ProbePaused };
+        string json = JsonSerializer.Serialize(heartbeat, MoShiroJsonContext.Default.AgentHeartbeatExtra);
+
+        Assert.Contains("\"runtime_status\":\"probe_paused\"", json, StringComparison.Ordinal);
+        Assert.Equal(NodeRuntimeStatus.ProbePaused,
+            JsonSerializer.Deserialize(json, MoShiroJsonContext.Default.AgentHeartbeatExtra)!.RuntimeStatus);
+    }
+}