Files
ClaudeDo/src/ClaudeDo.Worker/State/TaskStateService.cs
T
mika kuns a067a7bfff fix(mcp): report a Failed task's failureReason instead of a bare status
get_task/batch_get_tasks now return failureReason (max_turns|timeout|error|
cancelled|unknown) plus failureTurnsUsed/failureMaxTurns on a Failed task, so
max_turns (worktree usually fine, continue_task) is distinguishable from a
real error (reset_failed_task) without pulling get_task_log's raw NDJSON.
Classified and stamped onto TaskEntity by TaskRunner.MarkFailed via
TaskStateService.FailAsync; TaskRunEntity also keeps the CLI's raw
terminal_reason/result_subtype/errors for deeper diagnosis. reset_failed_task's
description now warns explicitly that it discards the worktree and points at
continue_task for max_turns. Surfaced on the task card's status-chip tooltip.
2026-08-10 14:20:00 +02:00

569 lines
26 KiB
C#

using ClaudeDo.Data;
using ClaudeDo.Data.Models;
using ClaudeDo.Data.Repositories;
using ClaudeDo.Worker.Hub;
using ClaudeDo.Worker.Planning;
using ClaudeDo.Worker.Queue;
using Microsoft.EntityFrameworkCore;
using TaskStatus = ClaudeDo.Data.Models.TaskStatus;
namespace ClaudeDo.Worker.State;
public sealed class TaskStateService : ITaskStateService
{
private readonly IDbContextFactory<ClaudeDoDbContext> _dbFactory;
private readonly HubBroadcaster _broadcaster;
private readonly IQueueWaker _waker;
private readonly PlanningChainCoordinator _chain;
private readonly RunCancellationRegistry _runCancels;
private readonly Func<IActiveMergeState> _mergeState;
private readonly ILogger<TaskStateService> _logger;
public TaskStateService(
IDbContextFactory<ClaudeDoDbContext> dbFactory,
HubBroadcaster broadcaster,
IQueueWaker waker,
PlanningChainCoordinator chain,
RunCancellationRegistry runCancels,
Func<IActiveMergeState> mergeState,
ILogger<TaskStateService> logger)
{
_dbFactory = dbFactory;
_broadcaster = broadcaster;
_waker = waker;
_chain = chain;
_runCancels = runCancels;
_mergeState = mergeState;
_logger = logger;
}
public async Task<TransitionResult> EnqueueAsync(string taskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
if (await IsDraftChildAsync(ctx, taskId, ct))
return new TransitionResult(false, "Draft subtask: finalize the plan before queuing it.");
if (await IsManualAsync(ctx, taskId, ct))
return new TransitionResult(false, "Manual task: mark it as a Claude task before queuing it.");
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status != TaskStatus.Running)
.ExecuteUpdateAsync(s => s.SetProperty(t => t.Status, TaskStatus.Queued), ct);
if (affected == 0)
return new TransitionResult(false, "Task not found or already running.");
_waker.Wake();
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> StartRunningAsync(string taskId, DateTime startedAt, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
if (await IsDraftChildAsync(ctx, taskId, ct))
return new TransitionResult(false, "Draft subtask: finalize the plan before running it.");
if (await IsManualAsync(ctx, taskId, ct))
return new TransitionResult(false, "Manual task: mark it as a Claude task before running it.");
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status != TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Running)
.SetProperty(t => t.StartedAt, startedAt), ct);
if (affected == 0)
return new TransitionResult(false, "Task already running or not found.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> CompleteAsync(string taskId, DateTime finishedAt, string? result, CancellationToken ct)
{
await using (var ctx = await _dbFactory.CreateDbContextAsync(ct))
{
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Done)
.SetProperty(t => t.FinishedAt, finishedAt)
.SetProperty(t => t.Result, result), ct);
if (affected == 0)
return new TransitionResult(false, "Task not running; cannot complete.");
}
await OnChildTerminalAsync(taskId, TaskStatus.Done);
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> SubmitForReviewAsync(string taskId, DateTime finishedAt, string? result, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.WaitingForReview)
.SetProperty(t => t.FinishedAt, finishedAt)
.SetProperty(t => t.Result, result), ct);
if (affected == 0)
return new TransitionResult(false, "Task not running; cannot submit for review.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
// Submit an interactively-worked task (a ConPTY session left its worktree with commits/changes)
// for review. Unlike SubmitForReviewAsync — which only fires from the headless Running state —
// this transitions from Idle or Failed, the states an interactive task sits in after the user
// finishes the session by hand. The caller commits the worktree first so there is a diff to merge.
public async Task<TransitionResult> SubmitInteractiveForReviewAsync(string taskId, DateTime finishedAt, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && (t.Status == TaskStatus.Idle || t.Status == TaskStatus.Failed))
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.WaitingForReview)
.SetProperty(t => t.FinishedAt, finishedAt), ct);
if (affected == 0)
return new TransitionResult(false, "Task is not Idle or Failed; cannot submit for review.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> SubmitForChildrenAsync(string taskId, DateTime finishedAt, string? result, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.WaitingForChildren)
.SetProperty(t => t.FinishedAt, finishedAt)
.SetProperty(t => t.Result, result), ct);
if (affected == 0)
return new TransitionResult(false, "Task not running; cannot submit for children.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> ApproveReviewAsync(string taskId, CancellationToken ct)
{
var now = DateTime.UtcNow;
await using (var ctx = await _dbFactory.CreateDbContextAsync(ct))
{
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.WaitingForReview)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Done)
.SetProperty(t => t.FinishedAt, now), ct);
if (affected == 0)
return new TransitionResult(false, "Task is not waiting for review; cannot approve.");
}
await OnChildTerminalAsync(taskId, TaskStatus.Done);
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> RejectToQueueAsync(string taskId, string feedback, CancellationToken ct)
{
if (string.IsNullOrWhiteSpace(feedback))
return new TransitionResult(false, "Feedback is required to reject for re-run.");
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.WaitingForReview)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Queued)
.SetProperty(t => t.ReviewFeedback, feedback)
.SetProperty(t => t.StartedAt, (DateTime?)null)
.SetProperty(t => t.FinishedAt, (DateTime?)null), ct);
if (affected == 0)
return new TransitionResult(false, "Task is not waiting for review; cannot reject.");
_waker.Wake();
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> RejectToIdleAsync(string taskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status == TaskStatus.WaitingForReview)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Idle)
.SetProperty(t => t.ReviewFeedback, (string?)null), ct);
if (affected == 0)
return new TransitionResult(false, "Task is not waiting for review; cannot park.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> ClearReviewFeedbackAsync(string taskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId)
.ExecuteUpdateAsync(s => s.SetProperty(t => t.ReviewFeedback, (string?)null), ct);
return affected == 0
? new TransitionResult(false, "Task not found.")
: new TransitionResult(true, null);
}
public async Task<TransitionResult> FailAsync(
string taskId, DateTime finishedAt, string? error, CancellationToken ct,
string failureReason = "error", int? turnsUsed = null, int? maxTurns = null)
{
await using (var ctx = await _dbFactory.CreateDbContextAsync(ct))
{
// Queued is intentional: OverrideSlotService dispatches RunAsync before calling
// StartRunningAsync, so a preflight failure (list not found, worktree setup) can
// reach MarkFailed while the task is still Queued in the DB.
var affected = await ctx.Tasks
.Where(t => t.Id == taskId &&
(t.Status == TaskStatus.Running || t.Status == TaskStatus.Queued))
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Failed)
.SetProperty(t => t.FinishedAt, finishedAt)
.SetProperty(t => t.Result, error)
.SetProperty(t => t.FailureReason, failureReason)
.SetProperty(t => t.FailureTurnsUsed, turnsUsed)
.SetProperty(t => t.FailureMaxTurns, maxTurns), ct);
if (affected == 0)
return new TransitionResult(false, "Task not in a failable state (must be Running or Queued).");
}
await OnChildTerminalAsync(taskId, TaskStatus.Failed);
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
// allowFromIdle: only the external update_task_status(Cancelled) path sets this — it lets
// an Idle task be retired without deleting it. Every other caller (hub CancelReview,
// PlanningChainCoordinator's chain-walk, batch cancel) relies on Idle staying a no-op here;
// PlanningChainCoordinator specifically uses "parked back to Idle" as a deliberate opt-out
// signal, so do not flip this default.
public async Task<TransitionResult> CancelAsync(
string taskId, DateTime finishedAt, CancellationToken ct, bool allowFromIdle = false)
{
// A unit merge drains its children's worktrees onto the target branch while the parent
// sits in WaitingForReview for the whole (potentially minutes-long) drain. Cancelling out
// from under it would flip the parent to Cancelled while the orchestrator keeps merging —
// FinalizeParentDoneAsync then finds the parent no longer WaitingForReview and gives up,
// leaving the merged children's diffs stranded with no way to roll them back.
if (_mergeState().HasActiveMerge(taskId))
return new TransitionResult(false, "A merge is in progress for this task; wait for it to finish before cancelling.");
List<string> cancelledChildIds;
await using (var ctx = await _dbFactory.CreateDbContextAsync(ct))
{
var affected = await ctx.Tasks
.Where(t => t.Id == taskId &&
(t.Status == TaskStatus.Running || t.Status == TaskStatus.Queued
|| t.Status == TaskStatus.WaitingForReview
|| t.Status == TaskStatus.WaitingForChildren
|| (allowFromIdle && t.Status == TaskStatus.Idle)))
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Cancelled)
.SetProperty(t => t.FinishedAt, finishedAt), ct);
if (affected == 0)
return new TransitionResult(false, "Task not in cancellable state.");
// Cascade to this task's own non-terminal children (one level — children are
// leaves in the current model) so a Queued/Running child doesn't keep going
// against a parent that's already gone, orphaning its worktree commit.
cancelledChildIds = await ctx.Tasks
.Where(t => t.ParentTaskId == taskId
&& t.Status != TaskStatus.Done
&& t.Status != TaskStatus.Failed
&& t.Status != TaskStatus.Cancelled)
.Select(t => t.Id)
.ToListAsync(ct);
if (cancelledChildIds.Count > 0)
await ctx.Tasks
.Where(t => cancelledChildIds.Contains(t.Id))
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Cancelled)
.SetProperty(t => t.FinishedAt, finishedAt)
.SetProperty(t => t.BlockedByTaskId, (string?)null), ct);
}
// Also stop the actual Claude processes: the DB flip above doesn't reach a run
// already executing in a queue/override slot — without this it would keep going
// and only fail its final (now-invalid) transition.
_runCancels.TryCancel(taskId);
foreach (var childId in cancelledChildIds)
_runCancels.TryCancel(childId);
await OnChildTerminalAsync(taskId, TaskStatus.Cancelled);
await _broadcaster.TaskUpdated(taskId);
foreach (var childId in cancelledChildIds)
await _broadcaster.TaskUpdated(childId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> ResetToIdleAsync(string taskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId && t.Status != TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Idle)
.SetProperty(t => t.StartedAt, (DateTime?)null)
.SetProperty(t => t.FinishedAt, (DateTime?)null)
.SetProperty(t => t.Result, (string?)null), ct);
if (affected == 0)
return new TransitionResult(false, "Task is running; cannot reset.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
// Unconditional status write — bypasses transition rules. Used by the UI's
// "set status freely" affordance; intentionally no guards (caller may strand
// the runner if used while a task is executing). It also bypasses chain/parent
// advancement — forcing a chain child or a WaitingForChildren parent's child to a
// terminal status here does not unblock its successor or re-check the parent.
public async Task<TransitionResult> ForceSetStatusAsync(string taskId, TaskStatus status, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = status == TaskStatus.Done
? await ctx.Tasks
.Where(t => t.Id == taskId)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, status)
.SetProperty(t => t.FinishedAt, DateTime.UtcNow), ct)
: await ctx.Tasks
.Where(t => t.Id == taskId)
.ExecuteUpdateAsync(s => s.SetProperty(t => t.Status, status), ct);
if (affected == 0)
return new TransitionResult(false, "Task not found.");
if (status == TaskStatus.Queued) _waker.Wake();
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> StartPlanningAsync(string parentId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == parentId
&& t.Status == TaskStatus.Idle
&& t.PlanningPhase == PlanningPhase.None)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.PlanningPhase, PlanningPhase.Active), ct);
if (affected == 0)
return new TransitionResult(false, "Task not in plannable state.");
await _broadcaster.TaskUpdated(parentId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> FinalizePlanningAsync(string parentId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var hasChildren = await ctx.Tasks.AnyAsync(t => t.ParentTaskId == parentId, ct);
var newStatus = hasChildren ? TaskStatus.WaitingForChildren : TaskStatus.WaitingForReview;
var affected = await ctx.Tasks
.Where(t => t.Id == parentId && t.PlanningPhase == PlanningPhase.Active)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.PlanningPhase, PlanningPhase.Finalized)
.SetProperty(t => t.PlanningFinalizedAt, DateTime.UtcNow)
.SetProperty(t => t.PlanningSessionToken, (string?)null)
.SetProperty(t => t.Status, newStatus), ct);
if (affected == 0)
return new TransitionResult(false, "No active planning session.");
await _broadcaster.TaskUpdated(parentId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> BlockOnAsync(string taskId, string predecessorTaskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId)
.ExecuteUpdateAsync(s => s.SetProperty(t => t.BlockedByTaskId, predecessorTaskId), ct);
if (affected == 0)
return new TransitionResult(false, "Task not found.");
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<TransitionResult> UnblockAsync(string taskId, CancellationToken ct)
{
await using var ctx = await _dbFactory.CreateDbContextAsync(ct);
var affected = await ctx.Tasks
.Where(t => t.Id == taskId)
.ExecuteUpdateAsync(s => s.SetProperty(t => t.BlockedByTaskId, (string?)null), ct);
if (affected == 0)
return new TransitionResult(false, "Task not found.");
_waker.Wake();
await _broadcaster.TaskUpdated(taskId);
return new TransitionResult(true, null);
}
public async Task<int> RecoverStaleRunningAsync(string reason, CancellationToken ct)
{
var resultText = "[stale] " + reason;
var now = DateTime.UtcNow;
List<string> recoveredIds;
int affected;
await using (var ctx = await _dbFactory.CreateDbContextAsync(ct))
{
recoveredIds = await ctx.Tasks
.Where(t => t.Status == TaskStatus.Running)
.Select(t => t.Id)
.ToListAsync(ct);
affected = await ctx.Tasks
.Where(t => t.Status == TaskStatus.Running)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.Failed)
.SetProperty(t => t.FinishedAt, now)
.SetProperty(t => t.Result, resultText), ct);
}
// A recovered task may have been a planning/improvement chain child or the last
// non-terminal child of a WaitingForChildren parent. The bulk flip above skips the
// usual terminal-transition side effects, so without this a crash mid-run would
// leave the chain successor blocked forever and the parent wedged in
// WaitingForChildren with nothing left to re-check it.
foreach (var taskId in recoveredIds)
await OnChildTerminalAsync(taskId, TaskStatus.Failed);
return affected;
}
// A subtask is "draft" only while its planning parent has an open (Active) session.
// Improvement children whose parent has PlanningPhase.None are not drafts and may be
// queued freely. Standalone tasks (no parent) are never draft.
// Server-side backstop for the manual flag: the UI hides the hand-off affordances, but the
// MCP surface and hub can still be driven directly.
private static Task<bool> IsManualAsync(ClaudeDoDbContext ctx, string taskId, CancellationToken ct)
=> ctx.Tasks.AsNoTracking().AnyAsync(t => t.Id == taskId && t.IsManual, ct);
private static async Task<bool> IsDraftChildAsync(ClaudeDoDbContext ctx, string taskId, CancellationToken ct)
{
var parentId = await ctx.Tasks.AsNoTracking()
.Where(t => t.Id == taskId)
.Select(t => t.ParentTaskId)
.FirstOrDefaultAsync(ct);
if (parentId is null) return false;
return await ctx.Tasks.AsNoTracking()
.AnyAsync(p => p.Id == parentId && p.PlanningPhase == PlanningPhase.Active, ct);
}
private async Task OnChildTerminalAsync(string taskId, TaskStatus finalStatus)
{
// Terminal child writes are best-effort and use CancellationToken.None so the
// task lifecycle is never left partially completed because a caller cancelled.
string? parentId;
await using (var ctx = await _dbFactory.CreateDbContextAsync(CancellationToken.None))
{
parentId = await ctx.Tasks
.AsNoTracking()
.Where(t => t.Id == taskId)
.Select(t => t.ParentTaskId)
.FirstOrDefaultAsync(CancellationToken.None);
}
if (parentId is null) return;
try
{
await _chain.OnChildFinishedAsync(taskId, finalStatus, CancellationToken.None);
}
catch (Exception ex)
{
_logger.LogWarning(ex, "PlanningChain advance failed for {TaskId}", taskId);
}
await TryAdvanceParentAsync(parentId);
}
// Any parent (planning or improvement) sitting in WaitingForChildren surfaces for review
// once every child is terminal (Done/Failed/Cancelled). A failed or cancelled child does
// not wedge the parent — it is flagged on the result. Also called directly after a child
// is deleted, since no terminal transition fires in that case.
public async Task TryAdvanceParentAsync(string parentId)
{
try
{
await AdvanceParentIfAllChildrenTerminalAsync(parentId);
}
catch (Exception ex)
{
_logger.LogWarning(ex, "TryAdvanceParent failed for {ParentId}", parentId);
}
}
private async Task AdvanceParentIfAllChildrenTerminalAsync(string parentId)
{
string? parentResult;
List<TaskStatus> childStatuses;
await using (var ctx = await _dbFactory.CreateDbContextAsync(CancellationToken.None))
{
var parent = await ctx.Tasks.AsNoTracking()
.FirstOrDefaultAsync(t => t.Id == parentId, CancellationToken.None);
if (parent is null || parent.Status != TaskStatus.WaitingForChildren) return;
parentResult = parent.Result;
childStatuses = await ctx.Tasks
.Where(t => t.ParentTaskId == parentId)
.Select(t => t.Status)
.ToListAsync(CancellationToken.None);
}
// No early-out on an empty list: zero children left (e.g. the last one was just
// deleted) counts as "all terminal" — .All() on an empty sequence is vacuously true.
bool allTerminal = childStatuses.All(s =>
s == TaskStatus.Done || s == TaskStatus.Failed || s == TaskStatus.Cancelled);
if (!allTerminal) return;
int failed = childStatuses.Count(s => s == TaskStatus.Failed);
int cancelled = childStatuses.Count(s => s == TaskStatus.Cancelled);
var newResult = parentResult;
if (failed + cancelled > 0)
{
var note = $"⚠ Children: {failed} failed, {cancelled} cancelled.";
newResult = string.IsNullOrWhiteSpace(parentResult) ? note : $"{parentResult}\n\n{note}";
}
await using var writeCtx = await _dbFactory.CreateDbContextAsync(CancellationToken.None);
await writeCtx.Tasks
.Where(t => t.Id == parentId && t.Status == TaskStatus.WaitingForChildren)
.ExecuteUpdateAsync(s => s
.SetProperty(t => t.Status, TaskStatus.WaitingForReview)
.SetProperty(t => t.Result, newResult), CancellationToken.None);
await _broadcaster.TaskUpdated(parentId);
}
}