- WP-1: Site cluster config (keep-oldest SBR, down-if-alone, 2s/10s failure detection) - WP-2: Site-role host bootstrap (no Kestrel, SQLite paths) - WP-3: SiteStorageService with deployed_configurations + static_attribute_overrides tables - WP-4: DeploymentManagerActor as cluster singleton with staggered Instance Actor creation, OneForOneStrategy/Resume supervision, deploy/disable/enable/delete lifecycle - WP-5: InstanceActor with attribute state, GetAttribute/SetAttribute, SQLite override persistence - WP-6: CoordinatedShutdown verified for graceful singleton handover - WP-7: Dual-node recovery (both seed nodes, min-nr-of-members=1) - WP-8: 31 tests (storage CRUD, actor lifecycle, supervision, negative checks) 389 total tests pass, zero warnings.
181 lines
6.6 KiB
C#
181 lines
6.6 KiB
C#
using Akka.Actor;
|
|
using Akka.Cluster;
|
|
using Akka.Cluster.Tools.Singleton;
|
|
using Akka.Configuration;
|
|
using Microsoft.Extensions.Options;
|
|
using ScadaLink.ClusterInfrastructure;
|
|
using ScadaLink.Host.Actors;
|
|
using ScadaLink.SiteRuntime;
|
|
using ScadaLink.SiteRuntime.Actors;
|
|
using ScadaLink.SiteRuntime.Persistence;
|
|
|
|
namespace ScadaLink.Host.Actors;
|
|
|
|
/// <summary>
|
|
/// Hosted service that manages the Akka.NET actor system lifecycle.
|
|
/// Creates the actor system on start, registers actors, and triggers
|
|
/// CoordinatedShutdown on stop.
|
|
/// </summary>
|
|
public class AkkaHostedService : IHostedService
|
|
{
|
|
private readonly IServiceProvider _serviceProvider;
|
|
private readonly NodeOptions _nodeOptions;
|
|
private readonly ClusterOptions _clusterOptions;
|
|
private readonly ILogger<AkkaHostedService> _logger;
|
|
private ActorSystem? _actorSystem;
|
|
|
|
public AkkaHostedService(
|
|
IServiceProvider serviceProvider,
|
|
IOptions<NodeOptions> nodeOptions,
|
|
IOptions<ClusterOptions> clusterOptions,
|
|
ILogger<AkkaHostedService> logger)
|
|
{
|
|
_serviceProvider = serviceProvider;
|
|
_nodeOptions = nodeOptions.Value;
|
|
_clusterOptions = clusterOptions.Value;
|
|
_logger = logger;
|
|
}
|
|
|
|
/// <summary>
|
|
/// Gets the actor system once started. Null before StartAsync completes.
|
|
/// </summary>
|
|
public ActorSystem? ActorSystem => _actorSystem;
|
|
|
|
public Task StartAsync(CancellationToken cancellationToken)
|
|
{
|
|
var seedNodesStr = string.Join(",",
|
|
_clusterOptions.SeedNodes.Select(s => $"\"{s}\""));
|
|
|
|
// For site nodes, include a site-specific role (e.g., "site-SiteA") alongside the base role
|
|
var roles = BuildRoles();
|
|
var rolesStr = string.Join(",", roles.Select(r => $"\"{r}\""));
|
|
|
|
var hocon = $@"
|
|
akka {{
|
|
actor {{
|
|
provider = cluster
|
|
}}
|
|
remote {{
|
|
dot-netty.tcp {{
|
|
hostname = ""{_nodeOptions.NodeHostname}""
|
|
port = {_nodeOptions.RemotingPort}
|
|
}}
|
|
}}
|
|
cluster {{
|
|
seed-nodes = [{seedNodesStr}]
|
|
roles = [{rolesStr}]
|
|
min-nr-of-members = {_clusterOptions.MinNrOfMembers}
|
|
split-brain-resolver {{
|
|
active-strategy = {_clusterOptions.SplitBrainResolverStrategy}
|
|
stable-after = {_clusterOptions.StableAfter.TotalSeconds:F0}s
|
|
keep-oldest {{
|
|
down-if-alone = on
|
|
}}
|
|
}}
|
|
failure-detector {{
|
|
heartbeat-interval = {_clusterOptions.HeartbeatInterval.TotalSeconds:F0}s
|
|
acceptable-heartbeat-pause = {_clusterOptions.FailureDetectionThreshold.TotalSeconds:F0}s
|
|
}}
|
|
run-coordinated-shutdown-when-down = on
|
|
}}
|
|
coordinated-shutdown {{
|
|
run-by-clr-shutdown-hook = on
|
|
}}
|
|
}}";
|
|
|
|
var config = ConfigurationFactory.ParseString(hocon);
|
|
_actorSystem = ActorSystem.Create("scadalink", config);
|
|
|
|
_logger.LogInformation(
|
|
"Akka.NET actor system 'scadalink' started. Role={Role}, Roles={Roles}, Hostname={Hostname}, Port={Port}",
|
|
_nodeOptions.Role,
|
|
string.Join(", ", roles),
|
|
_nodeOptions.NodeHostname,
|
|
_nodeOptions.RemotingPort);
|
|
|
|
// Register the dead letter monitor actor
|
|
var loggerFactory = _serviceProvider.GetRequiredService<ILoggerFactory>();
|
|
var dlmLogger = loggerFactory.CreateLogger<DeadLetterMonitorActor>();
|
|
_actorSystem.ActorOf(
|
|
Props.Create(() => new DeadLetterMonitorActor(dlmLogger)),
|
|
"dead-letter-monitor");
|
|
|
|
// For site nodes, register the Deployment Manager as a cluster singleton
|
|
if (_nodeOptions.Role.Equals("Site", StringComparison.OrdinalIgnoreCase))
|
|
{
|
|
RegisterSiteActors();
|
|
}
|
|
|
|
return Task.CompletedTask;
|
|
}
|
|
|
|
public async Task StopAsync(CancellationToken cancellationToken)
|
|
{
|
|
if (_actorSystem != null)
|
|
{
|
|
_logger.LogInformation("Shutting down Akka.NET actor system via CoordinatedShutdown...");
|
|
var shutdown = Akka.Actor.CoordinatedShutdown.Get(_actorSystem);
|
|
await shutdown.Run(Akka.Actor.CoordinatedShutdown.ClrExitReason.Instance);
|
|
_logger.LogInformation("Akka.NET actor system shutdown complete.");
|
|
}
|
|
}
|
|
|
|
/// <summary>
|
|
/// Builds the list of cluster roles for this node. Site nodes get both "Site"
|
|
/// and a site-specific role (e.g., "site-SiteA") to scope singleton placement.
|
|
/// </summary>
|
|
private List<string> BuildRoles()
|
|
{
|
|
var roles = new List<string> { _nodeOptions.Role };
|
|
|
|
if (_nodeOptions.Role.Equals("Site", StringComparison.OrdinalIgnoreCase)
|
|
&& !string.IsNullOrEmpty(_nodeOptions.SiteId))
|
|
{
|
|
roles.Add($"site-{_nodeOptions.SiteId}");
|
|
}
|
|
|
|
return roles;
|
|
}
|
|
|
|
/// <summary>
|
|
/// Registers site-specific actors including the Deployment Manager cluster singleton.
|
|
/// The singleton is scoped to the site-specific cluster role so it runs on exactly
|
|
/// one node within this site's cluster.
|
|
/// </summary>
|
|
private void RegisterSiteActors()
|
|
{
|
|
var siteRole = $"site-{_nodeOptions.SiteId}";
|
|
var storage = _serviceProvider.GetRequiredService<SiteStorageService>();
|
|
var siteRuntimeOptionsValue = _serviceProvider.GetService<IOptions<SiteRuntimeOptions>>()?.Value
|
|
?? new SiteRuntimeOptions();
|
|
var dmLogger = _serviceProvider.GetRequiredService<ILoggerFactory>()
|
|
.CreateLogger<DeploymentManagerActor>();
|
|
|
|
// Create the Deployment Manager as a cluster singleton
|
|
var singletonProps = ClusterSingletonManager.Props(
|
|
singletonProps: Props.Create(() => new DeploymentManagerActor(
|
|
storage,
|
|
siteRuntimeOptionsValue,
|
|
dmLogger)),
|
|
terminationMessage: PoisonPill.Instance,
|
|
settings: ClusterSingletonManagerSettings.Create(_actorSystem!)
|
|
.WithRole(siteRole)
|
|
.WithSingletonName("deployment-manager"));
|
|
|
|
_actorSystem!.ActorOf(singletonProps, "deployment-manager-singleton");
|
|
|
|
// Create a proxy for other actors to communicate with the singleton
|
|
var proxyProps = ClusterSingletonProxy.Props(
|
|
singletonManagerPath: "/user/deployment-manager-singleton",
|
|
settings: ClusterSingletonProxySettings.Create(_actorSystem)
|
|
.WithRole(siteRole)
|
|
.WithSingletonName("deployment-manager"));
|
|
|
|
_actorSystem.ActorOf(proxyProps, "deployment-manager-proxy");
|
|
|
|
_logger.LogInformation(
|
|
"Site actors registered. DeploymentManager singleton scoped to role={SiteRole}",
|
|
siteRole);
|
|
}
|
|
}
|