Files
junhong_cmp_fiber/internal/task/polling_carddata_handler.go
邹进 2fe7c1f8ac feat: 按运营商账户调度的分发器、卡任务执行器与模式切换
- 四个轮询处理器拆为步骤函数 Step(返回结果分类),旧 Asynq Handle 只做并发控制与原有重入队映射,行为不变;carrier 模式下旧 handler 直接结束
- 网关限流增加等待模式(ctx 标记),NSQ 消费者等名额到单步超时,超时按限流处理不计失败;记录限流等待次数
- 卡任务执行器:核对 token 并抢执行权,按配置步骤顺序串行执行到期检查(单步超时、总超时、panic 恢复、流量已触发评估时跳过套餐),执行中续期并 Touch,续期失败立即停止;收尾写回到期、连续失败、退避、延迟确认与告警
- 分发器:每秒按账户抢分发锁领取(执行上限按消费能力收紧、优先占比、按限流组剩余 QPS 估算),发布到 polling.card.{账户},发布失败释放;消费旧优先提示与手动触发;心跳与处理中标记到期计数
- NSQ 生产者与按账户动态订阅的消费者管理器,优雅退出宽限期
- 失败告警异步写 tb_polling_failure_alert(同卡同类型一条未处理,成功转已恢复),账户窗口内告警数超阈值产生账户级告警
- 模式监听:切 carrier 时旧分片调度停止出队,加锁初始化新结构后开始领取并订阅;回 legacy 时停止消费、重建旧分片队列;生命周期(API 与 Worker)在 carrier 模式写账户队列
2026-10-25 18:44:23 +09:00

155 lines
7.4 KiB
Go

package task
import (
"context"
"strings"
"time"
"github.com/hibiken/asynq"
"go.uber.org/zap"
cardapp "github.com/break/junhong_cmp_fiber/internal/application/cardobservation"
carddomain "github.com/break/junhong_cmp_fiber/internal/domain/cardobservation"
"github.com/break/junhong_cmp_fiber/internal/gateway"
"github.com/break/junhong_cmp_fiber/internal/infrastructure/integrationlog"
"github.com/break/junhong_cmp_fiber/internal/store/postgres"
"github.com/break/junhong_cmp_fiber/pkg/constants"
)
// PollingCarddataHandler 负责流量轮询编排,查询结果统一交给卡观测应用服务。
type PollingCarddataHandler struct {
base *PollingBase
gateway *gateway.Client
carrier *postgres.CarrierStore
observation *cardapp.Service
integration *integrationlog.Repository
}
// NewPollingCarddataHandler 创建流量检查任务处理器。
func NewPollingCarddataHandler(base *PollingBase, gw *gateway.Client, carrier *postgres.CarrierStore, observation *cardapp.Service, integration *integrationlog.Repository) *PollingCarddataHandler {
return &PollingCarddataHandler{base: base, gateway: gw, carrier: carrier, observation: observation, integration: integration}
}
// Handle 处理流量检查任务(旧分片调度)。按运营商账户调度时直接结束。
func (h *PollingCarddataHandler) Handle(ctx context.Context, task *asynq.Task) error {
cardID, ok := parseTaskPayload(task.Payload(), h.base.logger)
if !ok {
return nil
}
if h.base.carrierSchedulerActive(ctx) {
return nil
}
if !h.base.acquireConcurrency(ctx, constants.TaskTypePollingCarddata) {
h.base.logger.Debug("并发已满,重新入队", zap.Uint("card_id", cardID))
return h.base.requeueCard(ctx, cardID, constants.TaskTypePollingCarddata)
}
defer h.base.releaseConcurrency(ctx, constants.TaskTypePollingCarddata)
return h.base.applyLegacyOutcome(ctx, cardID, constants.TaskTypePollingCarddata, h.Step(ctx, cardID))
}
// Step 执行一次流量检查步骤,返回结果分类;不负责重入队。旧分片调度与按运营商账户调度共用。
func (h *PollingCarddataHandler) Step(ctx context.Context, cardID uint) StepOutcome {
startedAt := time.Now()
lockToken, locked, err := h.base.acquireCardTrafficSyncLock(ctx, cardID)
if err != nil || !locked {
h.base.logger.Warn("卡流量同步锁不可用,延迟重入队", zap.Uint("card_id", cardID), zap.Error(err))
return stepDefer(5 * time.Second)
}
defer h.base.releaseCardTrafficSyncLock(ctx, cardID, lockToken)
// 优先轮询互斥接缝:放在流量同步锁获取成功之后,避免认领后拿不到锁导致优先项空占。
priority, skip, priorityErr := h.base.beginPollingPriority(ctx, cardID, constants.TaskTypePollingCarddata)
if priorityErr != nil {
return h.fail(ctx, cardID, startedAt, "查询活动优先轮询项失败", priorityErr)
}
if skip {
h.base.logger.Debug("卡轮询优先项已被其它执行持有,本轮跳过并延后",
zap.Uint("card_id", cardID), zap.String("task_type", constants.TaskTypePollingCarddata))
return stepDefer(0)
}
h.base.invalidateCardCache(ctx, cardID)
card, err := h.base.iotCardStore.GetByID(ctx, cardID)
if err != nil {
if isNotFound(err) {
priority.FailFinal(ctx, "卡不存在")
return StepOutcome{Result: StepCardGone}
}
// 本地读卡失败属可恢复失败:先按真实执行累加尝试次数,再按上限收敛,避免加急事实被静默丢弃。
priority.MarkExecution(ctx)
priority.FailRetryable(ctx, "获取卡信息失败")
return h.fail(ctx, cardID, startedAt, "获取卡信息失败", err)
}
// 执行前轮询范围校验:优先通道不得重启已停止轮询的卡。
if !h.base.priorityInPollingScope(ctx, priority, card) {
return stepSkip()
}
if h.gateway == nil {
priority.FailFinal(ctx, "流量查询能力未配置")
return stepSkip()
}
priority.MarkExecution(ctx)
attempt := newGatewayAttempt(cardID, constants.IntegrationOperationGatewayTraffic, constants.CardObservationSceneTrafficPolling)
result, err := h.gateway.QueryFlow(gateway.WithCarrier(ctx, card.CarrierID), &gateway.FlowQueryReq{CardNo: card.ICCID})
if gateway.IsRateLimited(err) {
priority.FailRetryable(ctx, "运营商账户网关限流,稍后重试")
return StepOutcome{Result: StepRateLimited}
}
if err != nil {
_ = recordPriorityGatewayAttempt(ctx, h.integration, attempt, constants.IntegrationResultFailed, false, "查询流量失败", priority)
priority.FailRetryable(ctx, "查询流量失败")
return h.fail(ctx, cardID, startedAt, "查询流量失败", err)
}
if strings.TrimSpace(result.ICCID) == "" {
_ = recordPriorityGatewayAttempt(ctx, h.integration, attempt, constants.IntegrationResultInvalidPayload, false, "流量查询响应缺少 ICCID", priority)
priority.FailRetryable(ctx, "流量查询响应缺少 ICCID")
return h.fail(ctx, cardID, startedAt, "流量查询响应缺少 ICCID", nil)
}
ctx = withPollingWorkerAuditContext(ctx, constants.TaskTypePollingCarddata, "卡流量轮询任务", attempt.IntegrationID)
if h.observation == nil || h.carrier == nil {
priority.FailFinal(ctx, "卡流量观测能力未配置")
return h.fail(ctx, cardID, startedAt, "卡流量观测能力未配置", nil)
}
now := time.Now()
decision, err := h.observation.ApplyTrafficObservation(ctx, carddomain.TrafficObservation{
CardID: cardID, GatewayReadingMB: float64(result.Used), ResetDay: h.carrier.GetDataResetDay(ctx, card.CarrierID),
Metadata: carddomain.ObservationMetadata{
ObservationID: attempt.IntegrationID, Source: constants.CardObservationSourcePolling,
Scene: constants.CardObservationSceneTrafficPolling, ObservedAt: now, CorrelationID: attempt.IntegrationID,
},
})
if err != nil {
_ = recordPriorityGatewayAttempt(ctx, h.integration, attempt, constants.IntegrationResultUnknown, false, "应用流量观测失败", priority)
priority.FailRetryable(ctx, "应用流量观测失败")
return h.fail(ctx, cardID, startedAt, "应用流量观测失败", err)
}
h.base.configMgr.RecordPreResetCheck(ctx, card, now)
changed := decision.IncrementMB > 0 || decision.CrossMonth
if changed {
_ = recordPriorityGatewayAttempt(ctx, h.integration, attempt, constants.IntegrationResultSuccess, true, "流量业务事实发生变化", priority)
}
if h.base.verboseLog {
h.base.logger.Info("流量轮询详情", zap.Uint("card_id", cardID), zap.String("iccid", card.ICCID),
zap.Float64("gateway_flow_mb", float64(result.Used)), zap.Float64("increment_mb", decision.IncrementMB),
zap.Bool("is_cross_month", decision.CrossMonth), zap.Bool("reading_accepted", decision.ReadingAccepted))
}
metric := "polling.observation.unchanged"
if changed {
metric = "polling.observation.persisted"
}
h.base.logger.Info("流量轮询观测完成", zap.Uint("card_id", cardID), zap.Bool("changed", changed), zap.String("metric", metric))
priority.Complete(ctx)
h.base.updateStats(ctx, constants.TaskTypePollingCarddata, true, time.Since(startedAt))
return StepOutcome{Result: StepSucceeded, TriggeredEvaluation: changed}
}
func (h *PollingCarddataHandler) fail(ctx context.Context, cardID uint, startedAt time.Time, message string, err error) StepOutcome {
fields := []zap.Field{zap.Uint("card_id", cardID)}
if err != nil {
fields = append(fields, zap.Error(err))
}
h.base.logger.Warn(message, fields...)
h.base.updateStats(ctx, constants.TaskTypePollingCarddata, false, time.Since(startedAt))
return h.base.stepFail(ctx, constants.GatewayOpQueryFlow, message, err)
}