feat!: probes link to the panel; signed release updates for container nodes [CI]
ci/woodpecker/push/main-amd64-build Pipeline was successful
ci/woodpecker/push/main-amd64-build Pipeline was successful
Probes connect to the panel directly and get the agent list as desired state. CI signs the image digest of agent and probe releases, the panel forwards the manifest, and a host updater verifies it with a local public key, pulls by digest and rolls back when the new version is unhealthy.
This commit is contained in:
justme committed
1 parent
938ca1eb21
commit
993dd5c41c
57 files changed
+1640
-559
No files matched your search
@@ -0,0 +1,100 @@
|
||||
using System.Text;
|
||||
using System.Text.Json;
|
||||
using Fleet.Core;
|
||||
using Monitor.Protocol;
|
||||
|
||||
namespace Monitor.Measurement;
|
||||
|
||||
/// <summary>
|
||||
/// 节点进程和宿主机更新器之间的文件交换（容器里 <c>/etc/monitor</c>，也就是数据目录）：
|
||||
/// 节点把面板发来的清单和签名写进 <c>update/</c>，把「更新」请求写成 <c>update/apply.request</c>；更新器验签并更新，状态写在 root 拥有的 <c>update-status/status.json</c>。
|
||||
/// 节点进程不执行任何命令，也不信任清单的内容（验签在宿主机上）。没有安装更新器时目录写不了或没人读，调用只是返回 false。
|
||||
/// </summary>
|
||||
public sealed class UpdateBridge
|
||||
{
|
||||
private const int MaxManifestBytes = 4096;
|
||||
private const int MaxSignatureBytes = 512;
|
||||
private const int MaxStatusBytes = 4096;
|
||||
|
||||
private readonly string _updateDirectory;
|
||||
private readonly string _statusFile;
|
||||
private readonly Action<string, byte[]> _write;
|
||||
private readonly INodeLog _log;
|
||||
|
||||
/// <param name="write">写数据目录里的文件（相对路径，原子替换）：由节点传入 <c>ConfigStore.Write</c>——常驻进程里只有它可以写文件。</param>
|
||||
public UpdateBridge(string dataDirectory, Action<string, byte[]> write, INodeLog log)
|
||||
{
|
||||
_updateDirectory = Path.Combine(dataDirectory, "update");
|
||||
_statusFile = Path.Combine(dataDirectory, "update-status", "status.json");
|
||||
_write = write;
|
||||
_log = log;
|
||||
}
|
||||
|
||||
/// <summary>放下面板发来的清单和签名（和已有的一样就不动，免得每次重连都惊动更新器）。先写清单、后写签名：更新器监视的是签名文件。</summary>
|
||||
public bool WriteOffer(UpdateOfferParams offer)
|
||||
{
|
||||
byte[] manifest = Encoding.UTF8.GetBytes(offer.Manifest);
|
||||
byte[] signature = Encoding.UTF8.GetBytes(offer.Signature);
|
||||
if (manifest.Length is 0 or > MaxManifestBytes || signature.Length is 0 or > MaxSignatureBytes)
|
||||
{
|
||||
_log.Warn("ignoring an update offer of an unexpected size");
|
||||
return false;
|
||||
}
|
||||
|
||||
try
|
||||
{
|
||||
string manifestPath = Path.Combine(_updateDirectory, "offer.manifest");
|
||||
string signaturePath = Path.Combine(_updateDirectory, "offer.sig");
|
||||
if (Same(manifestPath, manifest) && Same(signaturePath, signature))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
_write("update/offer.manifest", manifest);
|
||||
_write("update/offer.sig", signature);
|
||||
_log.Info("update offer received");
|
||||
return true;
|
||||
}
|
||||
catch (Exception ex) when (ex is IOException or UnauthorizedAccessException)
|
||||
{
|
||||
_log.Warn("could not store the update offer (is the updater installed?)", ("err", ex.Message));
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>请求更新：写一个空的 <c>apply.request</c>（更新器验签通过、清单更新才会真的动手）。</summary>
|
||||
public bool RequestApply()
|
||||
{
|
||||
try
|
||||
{
|
||||
_write("update/apply.request", []);
|
||||
_log.Info("update requested");
|
||||
return true;
|
||||
}
|
||||
catch (Exception ex) when (ex is IOException or UnauthorizedAccessException)
|
||||
{
|
||||
_log.Warn("could not request the update (is the updater installed?)", ("err", ex.Message));
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>更新器的状态；没有安装更新器、文件不存在或内容不对时返回 null。</summary>
|
||||
public UpdateStatus? ReadStatus()
|
||||
{
|
||||
try
|
||||
{
|
||||
if (!File.Exists(_statusFile) || new FileInfo(_statusFile).Length > MaxStatusBytes)
|
||||
{
|
||||
return null;
|
||||
}
|
||||
|
||||
return JsonSerializer.Deserialize(File.ReadAllBytes(_statusFile), MonitorJsonContext.Default.UpdateStatus);
|
||||
}
|
||||
catch (Exception ex) when (ex is IOException or UnauthorizedAccessException or JsonException)
|
||||
{
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
private static bool Same(string path, byte[] content) => File.Exists(path) && File.ReadAllBytes(path).AsSpan().SequenceEqual(content);
|
||||
}
|
||||
@@ -13,8 +13,8 @@ public static class MonitorMethods
|
||||
public const string TestStatus = "monitor.test.status";
|
||||
public const string Event = "monitor.event";
|
||||
public const string Host = "monitor.host";
|
||||
public const string ProbeAgents = "monitor.probe.agents";
|
||||
public const string ProbeAgentsChanged = "monitor.probe.agents_changed";
|
||||
public const string UpdateOffer = "monitor.update.offer";
|
||||
public const string UpdateApply = "monitor.update.apply";
|
||||
|
||||
// Probe ↔ Agent（点对点会话）
|
||||
public const string Clock = "monitor.clock";
|
||||
@@ -147,19 +147,11 @@ public sealed class AgentEndpoint
|
||||
public string Spki { get; set; } = "";
|
||||
}
|
||||
|
||||
public sealed class ProbeAgentsParams
|
||||
/// <summary>期望状态文档（Fleet state type = <c>probe_agents</c>）：Backend 直接下发给每个 Probe 的 Agent 列表。</summary>
|
||||
public sealed class ProbeAgentList
|
||||
{
|
||||
/// <summary>Probe 目前持有的列表版本（用于调试；服务端总是返回当前列表）。</summary>
|
||||
[JsonPropertyName("have_version")]
|
||||
public long HaveVersion { get; set; }
|
||||
public const string StateType = "probe_agents";
|
||||
|
||||
/// <summary>Probe 的系统和架构（只用于展示；Agent 记在会话上，随连接状态报给 Backend）。</summary>
|
||||
[JsonPropertyName("platform")]
|
||||
public PlatformInfo? Platform { get; set; }
|
||||
}
|
||||
|
||||
public sealed class ProbeAgentsResult
|
||||
{
|
||||
[JsonPropertyName("version")]
|
||||
public long Version { get; set; }
|
||||
|
||||
@@ -227,6 +219,73 @@ public sealed class PlatformInfo
|
||||
public string? BuiltAt { get; set; }
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 面板给节点的更新清单（<c>monitor.update.offer</c>）：<c>manifest</c> 是发布清单的原始文本，<c>signature</c> 是 base64 的签名。
|
||||
/// 节点只是把它们原样放进数据目录；验签、拉取、重启都由宿主机上的更新器完成（节点进程不执行任何命令，也不信任清单的内容）。
|
||||
/// </summary>
|
||||
public sealed class UpdateOfferParams
|
||||
{
|
||||
[JsonPropertyName("manifest")]
|
||||
public string Manifest { get; set; } = "";
|
||||
|
||||
[JsonPropertyName("signature")]
|
||||
public string Signature { get; set; } = "";
|
||||
}
|
||||
|
||||
/// <summary>宿主机更新器已经验证过的可用更新。</summary>
|
||||
public sealed class UpdateAvailable
|
||||
{
|
||||
[JsonPropertyName("version")]
|
||||
public string? Version { get; set; }
|
||||
|
||||
[JsonPropertyName("tag")]
|
||||
public string? Tag { get; set; }
|
||||
|
||||
[JsonPropertyName("digest")]
|
||||
public string? Digest { get; set; }
|
||||
|
||||
[JsonPropertyName("seq")]
|
||||
public long Seq { get; set; }
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 宿主机更新器的状态（它写在 <c>update-status/status.json</c>，节点读出来随心跳上报）：
|
||||
/// <c>up_to_date</c>、<c>available</c>、<c>applying</c>、<c>failed</c>、<c>rolled_back</c>、<c>invalid</c>（清单验签或校验没过）。
|
||||
/// 没有安装更新器的节点没有这个字段。
|
||||
/// </summary>
|
||||
public sealed class UpdateStatus
|
||||
{
|
||||
[JsonPropertyName("state")]
|
||||
public string State { get; set; } = "";
|
||||
|
||||
/// <summary>正在运行的镜像 digest（更新器记录的；还没更新过时为空）。</summary>
|
||||
[JsonPropertyName("current")]
|
||||
public string? Current { get; set; }
|
||||
|
||||
[JsonPropertyName("available")]
|
||||
public UpdateAvailable? Available { get; set; }
|
||||
|
||||
[JsonPropertyName("error")]
|
||||
public string? Error { get; set; }
|
||||
|
||||
[JsonPropertyName("at")]
|
||||
public long At { get; set; }
|
||||
}
|
||||
|
||||
/// <summary>Probe 的 <c>fleet.heartbeat</c> 附加字段：更新器的状态。</summary>
|
||||
public sealed class ProbeHeartbeatExtra
|
||||
{
|
||||
[JsonPropertyName("update")]
|
||||
public UpdateStatus? Update { get; set; }
|
||||
}
|
||||
|
||||
/// <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
|
||||
{
|
||||
[JsonPropertyName("platform")]
|
||||
public PlatformInfo? Platform { get; set; }
|
||||
}
|
||||
|
||||
/// <summary><c>fleet.hello</c> 的附加字段。</summary>
|
||||
public sealed class AgentHelloMeta
|
||||
{
|
||||
@@ -302,9 +361,6 @@ public sealed class ConnectedProbe
|
||||
|
||||
[JsonPropertyName("since_utc_ms")]
|
||||
public long SinceUtcMs { get; set; }
|
||||
|
||||
[JsonPropertyName("platform")]
|
||||
public PlatformInfo? Platform { get; set; }
|
||||
}
|
||||
|
||||
/// <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>
|
||||
@@ -332,6 +388,9 @@ public sealed class AgentHeartbeatExtra
|
||||
|
||||
[JsonPropertyName("queue_bytes")]
|
||||
public long QueueBytes { get; set; }
|
||||
|
||||
[JsonPropertyName("update")]
|
||||
public UpdateStatus? Update { get; set; }
|
||||
}
|
||||
|
||||
/// <summary>Agent 下发给 Probe 的一条线路任务（期望状态类型 <c>line_task_set</c> 里的一项）。</summary>
|
||||
@@ -470,8 +529,11 @@ public static class FragmentKeys
|
||||
[JsonSerializable(typeof(HostSnapshot))]
|
||||
[JsonSerializable(typeof(TestStatusParams))]
|
||||
[JsonSerializable(typeof(AgentEventParams))]
|
||||
[JsonSerializable(typeof(ProbeAgentsParams))]
|
||||
[JsonSerializable(typeof(ProbeAgentsResult))]
|
||||
[JsonSerializable(typeof(ProbeAgentList))]
|
||||
[JsonSerializable(typeof(UpdateOfferParams))]
|
||||
[JsonSerializable(typeof(UpdateStatus))]
|
||||
[JsonSerializable(typeof(ProbeHeartbeatExtra))]
|
||||
[JsonSerializable(typeof(ProbeHelloMeta))]
|
||||
[JsonSerializable(typeof(AgentEnrollMeta))]
|
||||
[JsonSerializable(typeof(PlatformInfo))]
|
||||
[JsonSerializable(typeof(AgentHelloMeta))]
|
||||
|
||||
@@ -56,9 +56,6 @@ public sealed class TaskSet
|
||||
/// <summary>允许连接到这个 Agent 的 Probe 公钥。</summary>
|
||||
[JsonPropertyName("peers")]
|
||||
public List<PeerSpec> Peers { get; set; } = [];
|
||||
|
||||
[JsonPropertyName("probe_agents_version")]
|
||||
public long ProbeAgentsVersion { get; set; }
|
||||
}
|
||||
|
||||
public static class TaskSchedule
|
||||
|
||||
Reference in new issue
Block a user