6843c7efac
后端：
1.阶段 5 task 服务边界落地
- 新增 cmd/task 与 services/task/{dao,rpc,sv}，承载 task zrpc、tasks 表迁移和 task outbox 消费边界
- 新增 gateway/client/task、shared/contracts/task 和 task port，gateway /api/v1/task/* 切到 task zrpc client
- 将 task.urgency.promote.requested handler / relay / retry loop 迁入 cmd/task，单体 worker 不再消费 task outbox
- 保留单体 Agent 残留 task 查询的 publish-only 写入能力，避免迁移期 task 事件丢失
- active-scheduler task facts / due job scanner 切到 task RPC，并移除启动期 tasks 表依赖检查
- 更新阶段 5 文档，记录 task 切流点、旧实现保留、跨域 DB 依赖缩减和下一轮建议
- 补充 task rpc 示例配置
92 lines
3.3 KiB
Go
92 lines
3.3 KiB
Go
package sv
|
||
|
||
import (
|
||
"context"
|
||
"errors"
|
||
"fmt"
|
||
"log"
|
||
"time"
|
||
|
||
"github.com/LoveLosita/smartflow/backend/model"
|
||
"gorm.io/gorm"
|
||
)
|
||
|
||
// syncActiveScheduleJobBestEffort 在任务变更后同步主动调度 due job。
|
||
//
|
||
// 职责边界：
|
||
// 1. 只维护 important_urgent_task 的 job，不直接触发主动调度主链路；
|
||
// 2. 任务未完成且存在 urgency_threshold_at 时 upsert pending job；
|
||
// 3. 任务已完成或阈值为空时取消当前 pending job；
|
||
// 4. 当前任务接口尚未整体事务化，job 同步失败只记日志，避免任务主写入出现“已落库但接口失败”的更差体验。
|
||
func (ts *TaskService) syncActiveScheduleJobBestEffort(ctx context.Context, task *model.Task) {
|
||
if ts == nil || ts.activeScheduleDAO == nil || task == nil {
|
||
return
|
||
}
|
||
if task.IsCompleted || task.UrgencyThresholdAt == nil {
|
||
ts.cancelActiveScheduleJobBestEffort(ctx, task.UserID, task.ID, "task_not_schedulable")
|
||
return
|
||
}
|
||
|
||
job := &model.ActiveScheduleJob{
|
||
ID: activeScheduleJobID(task.UserID, task.ID),
|
||
UserID: task.UserID,
|
||
TaskID: task.ID,
|
||
TriggerType: model.ActiveScheduleTriggerTypeImportantUrgentTask,
|
||
Status: model.ActiveScheduleJobStatusPending,
|
||
TriggerAt: *task.UrgencyThresholdAt,
|
||
DedupeKey: activeScheduleTriggerDedupeKey(task.UserID, task.ID, *task.UrgencyThresholdAt),
|
||
TraceID: activeScheduleTraceID(task.UserID, task.ID),
|
||
}
|
||
if err := ts.activeScheduleDAO.CreateOrUpdateJob(ctx, job); err != nil {
|
||
log.Printf("主动调度 job upsert 失败: user_id=%d task_id=%d err=%v", task.UserID, task.ID, err)
|
||
}
|
||
}
|
||
|
||
// cancelActiveScheduleJobBestEffort 取消任务当前待触发 job。
|
||
//
|
||
// 职责边界：
|
||
// 1. 只取消 pending job，历史 triggered/skipped/failed 记录保留审计；
|
||
// 2. 找不到 pending job 属于正常幂等场景；
|
||
// 3. reason 只进入 last_error_code，方便后续排障知道取消来源。
|
||
func (ts *TaskService) cancelActiveScheduleJobBestEffort(ctx context.Context, userID int, taskID int, reason string) {
|
||
if ts == nil || ts.activeScheduleDAO == nil || userID <= 0 || taskID <= 0 {
|
||
return
|
||
}
|
||
job, err := ts.activeScheduleDAO.FindPendingJobByTask(ctx, userID, taskID)
|
||
if err != nil {
|
||
if errors.Is(err, gorm.ErrRecordNotFound) {
|
||
return
|
||
}
|
||
log.Printf("主动调度 pending job 查询失败: user_id=%d task_id=%d err=%v", userID, taskID, err)
|
||
return
|
||
}
|
||
now := time.Now()
|
||
updates := map[string]any{
|
||
"status": model.ActiveScheduleJobStatusCanceled,
|
||
"last_error_code": reason,
|
||
"last_scanned_at": &now,
|
||
}
|
||
if err = ts.activeScheduleDAO.UpdateJobFields(ctx, job.ID, updates); err != nil {
|
||
log.Printf("主动调度 pending job 取消失败: user_id=%d task_id=%d job_id=%s err=%v", userID, taskID, job.ID, err)
|
||
}
|
||
}
|
||
|
||
func activeScheduleJobID(userID int, taskID int) string {
|
||
return fmt.Sprintf("asj_task_%d_%d", userID, taskID)
|
||
}
|
||
|
||
func activeScheduleTraceID(userID int, taskID int) string {
|
||
return fmt.Sprintf("trace_active_task_%d_%d", userID, taskID)
|
||
}
|
||
|
||
func activeScheduleTriggerDedupeKey(userID int, taskID int, triggerAt time.Time) string {
|
||
windowStart := triggerAt.Truncate(30 * time.Minute)
|
||
return fmt.Sprintf("%d:%s:%s:%d:%s",
|
||
userID,
|
||
model.ActiveScheduleTriggerTypeImportantUrgentTask,
|
||
model.ActiveScheduleTargetTypeTaskPool,
|
||
taskID,
|
||
windowStart.Format(time.RFC3339),
|
||
)
|
||
}
|