[
https://issues.apache.org/jira/browse/GOBBLIN-1716?focusedWorklogId=813161&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-813161
]
ASF GitHub Bot logged work on GOBBLIN-1716:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 28/Sep/22 23:09
Start Date: 28/Sep/22 23:09
Worklog Time Spent: 10m
Work Description: ZihanLi58 commented on code in PR #3570:
URL: https://github.com/apache/gobblin/pull/3570#discussion_r982929378
##########
gobblin-runtime/src/main/java/org/apache/gobblin/runtime/kafka/HighLevelConsumer.java:
##########
@@ -123,19 +123,7 @@ public HighLevelConsumer(String topic, Config config, int
numThreads) {
this.numThreads = numThreads;
this.config = config.withFallback(FALLBACK);
this.gobblinKafkaConsumerClient = createConsumerClient(this.config);
- // On Partition rebalance, commit exisiting offsets and reset.
- this.gobblinKafkaConsumerClient.subscribe(this.topic, new
GobblinConsumerRebalanceListener() {
- @Override
- public void onPartitionsRevoked(Collection<KafkaPartition> partitions) {
- copyAndCommit();
- partitionOffsetsToCommit.clear();
- }
-
- @Override
- public void onPartitionsAssigned(Collection<KafkaPartition> partitions) {
- // No op
- }
- });
+ initializeConsumerClient();
Review Comment:
Change the method name to be assign topic partitions?
##########
gobblin-service/src/main/java/org/apache/gobblin/service/monitoring/SpecStoreChangeMonitor.java:
##########
@@ -74,7 +76,15 @@ public String load(String key) throws Exception {
protected GobblinServiceJobScheduler scheduler;
public SpecStoreChangeMonitor(String topic, Config config, int numThreads) {
- super(topic, config, numThreads);
+ // Differentiate group id for each host
+ super(topic, config.withValue(GROUP_ID_KEY,
+ ConfigValueFactory.fromAnyRef(SPEC_STORE_CHANGE_MONITOR_PREFIX +
UUID.randomUUID().toString())),
+ numThreads);
+ }
+
+ @Override
+ protected void initializeConsumerClient() {
Review Comment:
Add comment here to mention that we expect the client to consume all
partitions in the topic from the latest offset? You can even move that part of
logic here
##########
gobblin-modules/gobblin-kafka-common/src/main/java/org/apache/gobblin/kafka/client/GobblinKafkaConsumerClient.java:
##########
@@ -169,6 +169,8 @@ default long committed(KafkaPartition partition) {
public default void assignAndSeek(List<KafkaPartition> topicPartitions,
Map<KafkaPartition, LongWatermark> topicWatermarksMap) { return; }
+ public default void initializeClient(String topic) { return; }
Review Comment:
What's this method for?
Issue Time Tracking
-------------------
Worklog Id: (was: 813161)
Time Spent: 40m (was: 0.5h)
> refactor HighLevelConsumer to make consumer initialization configurable
> -----------------------------------------------------------------------
>
> Key: GOBBLIN-1716
> URL: https://issues.apache.org/jira/browse/GOBBLIN-1716
> Project: Apache Gobblin
> Issue Type: Improvement
> Reporter: Urmi Mustafi
> Priority: Major
> Time Spent: 40m
> Remaining Estimate: 0h
>
> Derived classes of HighLevelConsumer, for example SpecChangeMonitor, may want
> to initialize the kafka partitions and offset differently so we refactor the
> base class to grant this configurability. The default implementation remains
> the same as the original.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
