我正在尝试将项目从apache-storm迁移到twitter-heron。经过大量的努力,我能够摆脱大多数错误,例如使用className: "org.apache.storm.kafka.ZkHosts"
而不是className: "storm.kafka.ZkHosts"
。但是,我一直在提交拓扑。我使用Flux将拓扑提交给风暴。
我在ZkState中创建CuratorFramework对象时遇到NullPointerException。在进一步挖掘时,我发现an issue in github,如果未设置有关zookeeper的配置,则会导致此问题。
进一步调试我发现问题是因为我缺少ZkState.java:46中所需的以下配置。
storm.zookeeper.session.timeout
storm.zookeeper.connection.timeout
storm.zookeeper.retry.times
storm.zookeeper.retry.interval
虽然我已设法识别问题,但是,我不知道在我的配置中将其添加到何处。有人可以帮我在哪里添加上述配置。谢谢。
My Flux Config
name: "My_Topology"
components:
- id: "zkHosts"
className: "org.apache.storm.kafka.ZkHosts"
constructorArgs:
- "localhost:2181"
- id: "SpoutConfig"
className: "org.apache.storm.kafka.SpoutConfig"
constructorArgs:
- ref: "zkHosts" # brokerHosts
- "my-topic" # topic
- "/my-zkRoot" # zkRoot
- "my-id" # spoutId
properties:
- name: "zkServers"
value: ["localhost"]
- name: "zkPort"
value: 2181
- name: "zkRoot"
value: "/my-zkRoot"
- name: "retryInitialDelayMs"
value: 2000
- name: "retryDelayMultiplier"
value: 2
config:
topology.workers: 5
topology.testing.always.try.serialize: true
spouts:
- id: "kafka-spout"
className: "org.apache.storm.kafka.KafkaSpout"
parallelism: 1
constructorArgs:
- ref: "SpoutConfig"
bolts:
- id: "my-bolt"
className: "com.example.sample.MyBolt"
parallelism: 1
streams:
- name: "kafka_spout --> my_bolt"
from: "kafka-spout"
to: "my-bolt"
grouping:
type: SHUFFLE
答案 0 :(得分:1)
您可以将这些添加到您的Flux yaml文件的配置部分
config:
topology.workers: 5
topology.testing.always.try.serialize: true
storm.zookeeper.session.timeout: 30000
storm.zookeeper.connection.timeout: 30000
storm.zookeeper.retry.times: 5
storm.zookeeper.retry.interval: 2000
答案 1 :(得分:1)
我们已经将Storm Flux本地集成到Heron中以简化它。使用Heron ECO - 您可以编写两种不同API的拓扑
有关其他信息,请参阅此处的ECO文档
https://apache.github.io/incubator-heron/docs/developers/java/eco-api/