Knative Eventing Dead Letter Sink不触发问题排查及测试YAML求助

核心问题排查 #

你遇到的问题是两处配置错误导致DLQ规则未生效：

• 投递规则缩进错误：你提供的Subscription配置中，retry、backoffDelay、backoffPolicy三个参数错误写在了deadLetterSink的子层级下，正确结构是这三个参数和deadLetterSink平级，同属于spec.delivery的子字段。该错误会导致Knative无法识别重试+DLQ规则，仅走默认投递逻辑，不会触发DLQ。
• Sequence场景DLQ配置位置错误：Broker层面配置的DLQ仅负责「Broker到Sequence入口」这一步的投递失败兜底，Sequence内部每一步的服务执行失败，由Sequence自动创建的内部Subscription负责投递，默认不会继承Broker的DLQ规则。Knative v0.26及以上版本支持直接在Sequence.spec.delivery字段配置全局DLQ规则，更低版本需要给Sequence用到的每个Channel单独配置delivery规则。

服务返回非2xx状态码本身符合DLQ触发条件，你之前的场景未触发完全是配置错误导致。

最简可运行DLQ测试YAML #

1. 部署测试服务和DLQ接收服务 #

# 返回500错误的测试服务
apiVersion: serving.knative.dev/v1
kind: Service
metadata:
  name: cetf
  namespace: kafka
spec:
  template:
    spec:
      containers:
      - image: gcr.io/knative-samples/knative-route-demo:blue
        env:
        - name: ERROR_CODE
          value: "500"
---
# DLQ消息接收服务，会在日志中打印收到的云事件
apiVersion: serving.knative.dev/v1
kind: Service
metadata:
  name: dlq
  namespace: kafka
spec:
  template:
    spec:
      containers:
      - image: gcr.io/knative-releases/knative.dev/eventing/cmd/event_display:latest

2. 正确配置的KafkaChannel + Subscription #

apiVersion: messaging.knative.dev/v1beta1
kind: KafkaChannel
metadata:
  name: test-channel
  namespace: kafka
spec:
  numPartitions: 3
  replicationFactor: 1
---
apiVersion: messaging.knative.dev/v1
kind: Subscription
metadata:
  name: test-sub
  namespace: kafka
spec:
  channel:
    apiVersion: messaging.knative.dev/v1beta1
    kind: KafkaChannel
    name: test-channel
  # 注意delivery层级结构，重试参数和deadLetterSink平级
  delivery:
    retry: 1
    backoffPolicy: linear
    backoffDelay: PT1S
    deadLetterSink:
      ref:
        apiVersion: serving.knative.dev/v1
        kind: Service
        name: dlq
        namespace: kafka
  subscriber:
    ref:
      apiVersion: serving.knative.dev/v1
      kind: Service
      name: cetf

3. 支持DLQ的Sequence配置示例（Knative v0.26+） #

apiVersion: flows.knative.dev/v1
kind: Sequence
metadata:
  name: test-sequence
  namespace: kafka
spec:
  steps:
  - ref:
      apiVersion: serving.knative.dev/v1
      kind: Service
      name: cetf # 返回500的测试步骤
  - ref:
      apiVersion: serving.knative.dev/v1
      kind: Service
      name: normal-service # 正常处理服务
  # 全局配置Sequence所有步骤的DLQ规则
  delivery:
    retry: 1
    backoffPolicy: linear
    backoffDelay: PT1S
    deadLetterSink:
      ref:
        apiVersion: serving.knative.dev/v1
        kind: Service
        name: dlq
        namespace: kafka

验证方法 #

给测试Channel或者Broker发送云事件后，查看dlq服务的Pod日志，即可看到投递失败的消息已经被转发到DLQ。

内容的提问来源于stack exchange，提问作者Leeish