feat(worker): surface startup recovery via OperationProgress channel

The six Lifecycle/*Recovery hosted services now broadcast one
OperationProgress("startup-recovery", <phase>, current, total) message each
after they finish, instead of leaving the UI on a bare "connecting" text
during worker startup. IslandsShellViewModel subscribes and swaps in
"Recovering... (i/n)" (existing ops.worker.startupRecovery key, no locale
changes) while Worker.IsReconnecting is true, and clears it once actually
connected so a later transient reconnect doesn't replay stale text.

OperationProgress broadcasts to Clients.All with no replay-on-connect, so a
UI that hasn't finished its SignalR handshake yet can miss some or all of
these messages and simply keep showing "connecting" as before -- accepted
rather than adding a cached-state + reconnect-replay path (mirroring
RefreshExternalMergeConflictsAsync) for what is a fast, best-effort,
local-only startup sweep with no UI-visible failure mode beyond that.
This commit is contained in:
mika kuns
2026-08-21 13:36:25 +02:00
parent dcda067b48
commit 6ec81f653b
16 changed files with 254 additions and 11 deletions
@@ -20,14 +20,24 @@ public sealed partial class IslandsShellViewModel : ViewModelBase, IDisposable
public ListsIslandViewModel? Lists { get; }
public TasksIslandViewModel? Tasks { get; }
public DetailsIslandViewModel? Details { get; }
public IWorkerClient? Worker { get; }
public IWorkerClient? Worker { get; internal set; }
public MissionControlViewModel? MissionControl { get; }
public UsagePillViewModel? UsagePill { get; }
public UpdateCheckService UpdateCheck => _updateCheck;
// Stable opKey the six Lifecycle/*Recovery startup sweeps broadcast under (see
// OperationProgressOpKeys.StartupRecovery on the worker side — not shared across the
// SignalR wire, so the literal is duplicated here on purpose).
private const string StartupRecoveryOpKey = "startup-recovery";
// Last startup-recovery text received this connection cycle, shown in place of the generic
// "connecting" label. Cleared once actually connected so a later transient reconnect (not a
// fresh worker boot) doesn't replay stale recovery text.
private string? _startupRecoveryText;
public string ConnectionText =>
Worker?.IsConnected == true ? Loc.T("vm.connection.online")
: Worker?.IsReconnecting == true ? Loc.T("vm.connection.connecting")
: Worker?.IsReconnecting == true ? (_startupRecoveryText ?? Loc.T("vm.connection.connecting"))
: Loc.T("vm.connection.offline");
public bool IsOffline => Worker?.IsConnected != true && Worker?.IsReconnecting != true;
@@ -228,6 +238,18 @@ public sealed partial class IslandsShellViewModel : ViewModelBase, IDisposable
}
public void OnPlanningMergeCompleted(string planningTaskId) => ClearExternalMergeConflict(planningTaskId);
// Wired to Worker.OperationProgressEvent; also called directly by tests. A foreign opKey
// (merge phases, worktree cleanup, ...) is left untouched — this only tracks the
// startup-recovery channel.
public void OnOperationProgress(string opKey, string phase, int current, int total)
{
if (opKey != StartupRecoveryOpKey) return;
_startupRecoveryText = total > 0
? $"{Loc.T("ops.worker.startupRecovery")} ({current}/{total})"
: Loc.T("ops.worker.startupRecovery");
OnPropertyChanged(nameof(ConnectionText));
}
private void ClearExternalMergeConflict(string planningTaskId)
{
_externalMergeConflicts.Remove(planningTaskId);
@@ -358,11 +380,13 @@ public sealed partial class IslandsShellViewModel : ViewModelBase, IDisposable
{
if (e.PropertyName is nameof(IWorkerClient.IsConnected) or nameof(IWorkerClient.IsReconnecting))
{
if (Worker.IsConnected) _startupRecoveryText = null;
OnPropertyChanged(nameof(ConnectionText));
OnPropertyChanged(nameof(IsOffline));
OnPropertyChanged(nameof(CanOpenWorkerConnectionHelp));
}
};
Worker.OperationProgressEvent += OnOperationProgress;
Worker.WorkerLogReceivedEvent += OnWorkerLogReceived;
Worker.ConnectionRestoredEvent += () => _ = RefreshStaleWorkerCheckAsync();
Worker.PlanningMergeConflictEvent += OnPlanningMergeConflict;
@@ -410,6 +434,7 @@ public sealed partial class IslandsShellViewModel : ViewModelBase, IDisposable
public void Dispose()
{
if (Worker is not null) Worker.OperationProgressEvent -= OnOperationProgress;
_clearTimer.Stop();
_clearTimer.Dispose();
_connectTimer.Stop();
@@ -0,0 +1,8 @@
namespace ClaudeDo.Worker.Hub;
/// <summary>Stable, non-task-bound opKeys for <see cref="HubBroadcaster.OperationProgress"/>.</summary>
public static class OperationProgressOpKeys
{
/// <summary>The six startup-only <c>Lifecycle/*Recovery</c> hosted services.</summary>
public const string StartupRecovery = "startup-recovery";
}
@@ -1,4 +1,5 @@
using ClaudeDo.Data;
using ClaudeDo.Worker.Hub;
using Microsoft.EntityFrameworkCore;
namespace ClaudeDo.Worker.Lifecycle;
@@ -8,17 +9,22 @@ namespace ClaudeDo.Worker.Lifecycle;
/// </summary>
public sealed class AttachmentOrphanRecovery : IHostedService
{
public const string Phase = "attachments";
private readonly IDbContextFactory<ClaudeDoDbContext> _dbFactory;
private readonly AttachmentStore _store;
private readonly HubBroadcaster _broadcaster;
private readonly ILogger<AttachmentOrphanRecovery> _logger;
public AttachmentOrphanRecovery(
IDbContextFactory<ClaudeDoDbContext> dbFactory,
AttachmentStore store,
HubBroadcaster broadcaster,
ILogger<AttachmentOrphanRecovery> logger)
{
_dbFactory = dbFactory;
_store = store;
_broadcaster = broadcaster;
_logger = logger;
}
@@ -28,6 +34,7 @@ public sealed class AttachmentOrphanRecovery : IHostedService
if (taskIds.Count == 0)
{
_logger.LogInformation("Attachment orphan recovery: no attachment directories found");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
return;
}
@@ -45,6 +52,8 @@ public sealed class AttachmentOrphanRecovery : IHostedService
_logger.LogWarning("Attachment orphan recovery: removed {Count} orphaned attachment director(ies)", orphans.Count);
else
_logger.LogInformation("Attachment orphan recovery: no orphaned attachment directories found");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, taskIds.Count, taskIds.Count);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
@@ -1,6 +1,7 @@
using ClaudeDo.Data;
using ClaudeDo.Data.Models;
using ClaudeDo.Data.Repositories;
using ClaudeDo.Worker.Hub;
using Microsoft.EntityFrameworkCore;
namespace ClaudeDo.Worker.Lifecycle;
@@ -13,16 +14,21 @@ namespace ClaudeDo.Worker.Lifecycle;
/// </summary>
public sealed class LegacyWorktreeFolderRecovery : IHostedService
{
public const string Phase = "legacy-worktrees";
private const string LegacyFolderName = ".claudedo-worktrees";
private readonly IDbContextFactory<ClaudeDoDbContext> _dbFactory;
private readonly HubBroadcaster _broadcaster;
private readonly ILogger<LegacyWorktreeFolderRecovery> _logger;
public LegacyWorktreeFolderRecovery(
IDbContextFactory<ClaudeDoDbContext> dbFactory,
HubBroadcaster broadcaster,
ILogger<LegacyWorktreeFolderRecovery> logger)
{
_dbFactory = dbFactory;
_broadcaster = broadcaster;
_logger = logger;
}
@@ -40,6 +46,8 @@ public sealed class LegacyWorktreeFolderRecovery : IHostedService
LegacyFolderName, affected.Count, string.Join(", ", affected));
else
_logger.LogInformation("Legacy worktree folder recovery: no lists affected");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, lists.Count, lists.Count);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
@@ -1,5 +1,6 @@
using ClaudeDo.Data;
using ClaudeDo.Data.Repositories;
using ClaudeDo.Worker.Hub;
using Microsoft.EntityFrameworkCore;
namespace ClaudeDo.Worker.Lifecycle;
@@ -12,14 +13,19 @@ namespace ClaudeDo.Worker.Lifecycle;
/// </summary>
public sealed class OrphanRecovery : IHostedService
{
public const string Phase = "orphaned-children";
private readonly IDbContextFactory<ClaudeDoDbContext> _dbFactory;
private readonly HubBroadcaster _broadcaster;
private readonly ILogger<OrphanRecovery> _logger;
public OrphanRecovery(
IDbContextFactory<ClaudeDoDbContext> dbFactory,
HubBroadcaster broadcaster,
ILogger<OrphanRecovery> logger)
{
_dbFactory = dbFactory;
_broadcaster = broadcaster;
_logger = logger;
}
@@ -32,6 +38,8 @@ public sealed class OrphanRecovery : IHostedService
_logger.LogWarning("Orphan recovery: dequeued {Count} stuck child task(s)", dequeued);
else
_logger.LogInformation("Orphan recovery: no stuck child tasks found");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
@@ -1,5 +1,6 @@
using ClaudeDo.Data;
using ClaudeDo.Data.Repositories;
using ClaudeDo.Worker.Hub;
using Microsoft.EntityFrameworkCore;
namespace ClaudeDo.Worker.Lifecycle;
@@ -13,17 +14,22 @@ namespace ClaudeDo.Worker.Lifecycle;
/// </summary>
public sealed class PlanningLineageRecovery : IHostedService
{
public const string Phase = "planning-lineage";
private readonly IDbContextFactory<ClaudeDoDbContext> _dbFactory;
private readonly string _sessionsRoot;
private readonly HubBroadcaster _broadcaster;
private readonly ILogger<PlanningLineageRecovery> _logger;
public PlanningLineageRecovery(
IDbContextFactory<ClaudeDoDbContext> dbFactory,
string sessionsRoot,
HubBroadcaster broadcaster,
ILogger<PlanningLineageRecovery> logger)
{
_dbFactory = dbFactory;
_sessionsRoot = sessionsRoot;
_broadcaster = broadcaster;
_logger = logger;
}
@@ -32,6 +38,7 @@ public sealed class PlanningLineageRecovery : IHostedService
if (!Directory.Exists(_sessionsRoot))
{
_logger.LogInformation("Planning lineage recovery: sessions directory missing, nothing to scan");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
return;
}
@@ -39,6 +46,7 @@ public sealed class PlanningLineageRecovery : IHostedService
if (folders.Length == 0)
{
_logger.LogInformation("Planning lineage recovery: no session folders");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
return;
}
@@ -63,6 +71,8 @@ public sealed class PlanningLineageRecovery : IHostedService
if (restored == 0)
_logger.LogInformation("Planning lineage recovery: no candidates");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, folders.Length, folders.Length);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
@@ -1,4 +1,5 @@
using ClaudeDo.Data;
using ClaudeDo.Worker.Hub;
namespace ClaudeDo.Worker.Lifecycle;
@@ -10,16 +11,20 @@ namespace ClaudeDo.Worker.Lifecycle;
/// </summary>
public sealed class PromptFileRecovery : IHostedService
{
public const string Phase = "prompt-files";
private readonly ILogger<PromptFileRecovery> _logger;
private readonly HubBroadcaster _broadcaster;
private readonly string? _root;
public PromptFileRecovery(ILogger<PromptFileRecovery> logger, string? root = null)
public PromptFileRecovery(ILogger<PromptFileRecovery> logger, HubBroadcaster broadcaster, string? root = null)
{
_logger = logger;
_broadcaster = broadcaster;
_root = root;
}
public Task StartAsync(CancellationToken cancellationToken)
public async Task StartAsync(CancellationToken cancellationToken)
{
PromptFiles.ReconcileStaleDefaults(_root);
@@ -27,7 +32,8 @@ public sealed class PromptFileRecovery : IHostedService
if (orphans.Count == 0)
{
_logger.LogInformation("Prompt file recovery: no orphaned prompt files found");
return Task.CompletedTask;
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
return;
}
foreach (var orphan_path in orphans)
@@ -40,7 +46,7 @@ public sealed class PromptFileRecovery : IHostedService
orphan_path);
}
return Task.CompletedTask;
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, orphans.Count, orphans.Count);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
@@ -1,15 +1,20 @@
using ClaudeDo.Worker.Hub;
using ClaudeDo.Worker.State;
namespace ClaudeDo.Worker.Lifecycle;
public sealed class StaleTaskRecovery : IHostedService
{
public const string Phase = "stale-tasks";
private readonly ITaskStateService _state;
private readonly HubBroadcaster _broadcaster;
private readonly ILogger<StaleTaskRecovery> _logger;
public StaleTaskRecovery(ITaskStateService state, ILogger<StaleTaskRecovery> logger)
public StaleTaskRecovery(ITaskStateService state, HubBroadcaster broadcaster, ILogger<StaleTaskRecovery> logger)
{
_state = state;
_broadcaster = broadcaster;
_logger = logger;
}
@@ -20,6 +25,8 @@ public sealed class StaleTaskRecovery : IHostedService
_logger.LogWarning("Stale task recovery: flipped {Count} running task(s) to failed", flipped);
else
_logger.LogInformation("Stale task recovery: no stale tasks found");
await _broadcaster.OperationProgress(OperationProgressOpKeys.StartupRecovery, Phase, 0, 0);
}
public Task StopAsync(CancellationToken cancellationToken) => Task.CompletedTask;
+1
View File
@@ -174,6 +174,7 @@ builder.Services.AddSingleton(sp =>
builder.Services.AddHostedService(sp => new PlanningLineageRecovery(
sp.GetRequiredService<IDbContextFactory<ClaudeDoDbContext>>(),
planningSessionsDir,
sp.GetRequiredService<HubBroadcaster>(),
sp.GetRequiredService<ILogger<PlanningLineageRecovery>>()));
builder.Services.AddSingleton<ITerminalLauncher>(sp =>
new WindowsTerminalLauncher("wt.exe", cfg.ClaudeBin));