Been experimenting with switching a Storm 1.0.6 topology to Heron. Taking a baby step by removing all but the Kafka spout to see how things go. Have a main method as follows (modified from the original Flux version):
import org.apache.heron.eco.Eco;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
public class KafkaTopology {
public static void main(String[] args) throws Exception {
List<String> argList = new ArrayList<String>(Arrays.asList(args));
String file = KafkaTopology.class.getClassLoader().getResource("topology.yaml").getFile();
argList.add("local");
argList.add("--eco-config-file");
argList.add(file);
file = KafkaTopology.class.getClassLoader().getResource("dev.properties").getFile();
argList.add("--props");
argList.add(file);
argList.add("--sleep");
argList.add("36000000");
String[] ecoArgs = argList.toArray(new String[argList.size()]);
Eco.main(ecoArgs);
}
}
YAML is this:
name: "kafkaTopology-XXX_topologyVersion_XXX"
type: "storm"
config:
topology.workers: ${workers.config}
topology.max.spout.pending: ${max.spout.pending}
topology.message.timeout.secs: 120
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
properties:
kafka.mapper.zkServers: ${kafka.mapper.zkServers}
kafka.mapper.zkPort: ${kafka.mapper.zkPort}
bootstrap.servers: ${bootstrap.servers}
kafka.mapper.brokerZkStr: ${kafka.mapper.brokerZkStr}
kafka.topic.name: ${kafka.topic.name}
components:
- id: "zkHosts"
className: "org.apache.storm.kafka.ZkHosts"
constructorArgs:
- ${kafka.mapper.brokerZkStr}
- id: "rawMessageAndMetadataScheme"
className: "org.acme.storm.spout.RawMessageAndMetadataScheme"
- id: "messageMetadataSchemeAsMultiScheme"
className: "org.apache.storm.kafka.MessageMetadataSchemeAsMultiScheme"
constructorArgs:
- ref: "rawMessageAndMetadataScheme"
- id: "kafkaSpoutConfig"
className: "org.apache.storm.kafka.SpoutConfig"
constructorArgs:
# brokerHosts
- ref: "zkHosts"
# topic
- ${kafka.topic.name}
# zkRoot
- "/zkRootKafka.kafkaSpout.builder"
# id
- ${kafka.topic.name}
properties:
- name: "scheme"
ref: "messageMetadataSchemeAsMultiScheme"
- name: zkServers
value: ${kafka.mapper.zkServers}
- name: zkPort
value: ${kafka.mapper.zkPort}
# Retry Properties
- name: "retryInitialDelayMs"
value: 60000
- name: "retryDelayMultiplier"
value: 1.5
- name: "retryDelayMaxMs"
value: 14400000
- name: "retryLimit"
value: 0
# spout definitions
spouts:
- id: "kafka-spout"
className: "org.apache.storm.kafka.KafkaSpout"
parallelism: ${kafka.spout.parallelism}
constructorArgs:
- ref: "kafkaSpoutConfig"
Relevant POM entries:
<dependency>
<groupId>org.apache.heron</groupId>
<artifactId>heron-api</artifactId>
<version>0.20.3-incubating</version>
</dependency>
<dependency>
<groupId>org.apache.heron</groupId>
<artifactId>heron-storm</artifactId>
<version>0.20.3-incubating</version>
</dependency>
<dependency>
<groupId>org.apache.storm</groupId>
<artifactId>storm-kafka</artifactId>
<version>1.0.6</version>
</dependency>
Main method seems to run fine:
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.parser.EcoParser loadTopologyFromYaml
INFO: Parsing eco config file
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.parser.EcoParser loadTopologyFromYaml
INFO: Performing property substitution.
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.parser.EcoParser loadTopologyFromYaml
INFO: Performing environment variable substitution.
topology type is Storm
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.builder.storm.EcoBuilder buildConfig
INFO: Building topology config
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: ---------- TOPOLOGY DETAILS ----------
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: Topology Name: kafkaTopology-XXX_topologyVersion_XXX
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: --------------- SPOUTS ---------------
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: kafka-spout [1] (org.apache.storm.kafka.KafkaSpout)
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: ---------------- BOLTS ---------------
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: --------------- STREAMS ---------------
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.Eco printTopologyInfo
INFO: --------------------------------------
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.builder.storm.EcoBuilder buildTopologyBuilder
INFO: Building components
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.builder.storm.EcoBuilder buildTopologyBuilder
INFO: Building spouts
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.builder.storm.EcoBuilder buildTopologyBuilder
INFO: Building bolts
Apr 30, 2021 4:38:49 PM org.apache.heron.eco.builder.storm.EcoBuilder buildTopologyBuilder
INFO: Building streams
Process finished with exit code 0
Question 1: The topology exits immediately, is there an Eco flag equivalent to Flux '--sleep' to keep it running for a while (to debug, etc.)?
Question 2: Was a little surprised that I needed to pull storm-kafka in (thought there would be a Heron equivalent) - is this correct (or some other artifact?) and if so, is 1.0.6 an OK version to use or does Heron work better with another version?
Question 3: The above was with type: "storm"
in the YAML, trying type: "heron"
gives the following error:
INFO: Building spouts
Exception in thread "main" java.lang.ClassCastException: class org.apache.storm.kafka.KafkaSpout cannot be cast to class org.apache.heron.api.spout.IRichSpout (org.apache.storm.kafka.KafkaSpout and org.apache.heron.api.spout.IRichSpout are in unnamed module of loader 'app')
at org.apache.heron.eco.builder.heron.SpoutBuilder.buildSpouts(SpoutBuilder.java:42)
at org.apache.heron.eco.builder.heron.EcoBuilder.buildTopologyBuilder(EcoBuilder.java:70)
at org.apache.heron.eco.Eco.submit(Eco.java:125)
at org.apache.heron.eco.Eco.main(Eco.java:161)
at KafkaTopology.main(KafkaTopology.java:26)
Process finished with exit code 1
Is this just the way it is using Kafka, type needs to be storm and not heron, or is there some workaround here?