Wednesday, October 17, 2018
Druid indexing problem: Offsets out of range with no configured reset policy
I have the following set up:
1) kafka server
2) druid server (historical, broker, coordinator, overlord, middleManager, tranquility, zookeeper)
To get data from kafka I use kafka indexing service (supervisor), which is managed by overlord.
The problem is that recently I started seeing these errors and no data published to dataset:
After some research I found that I need to set this property in kafka consumer properties
Unfortunately I didn't find how to modify it in kafka, because on kafka server I run only kafka server like this:
But I found another way how to fix this problem. You can also specify one setting in supervisor config. Here is what you need to set:
and resubmit it to overlord:
1) kafka server
2) druid server (historical, broker, coordinator, overlord, middleManager, tranquility, zookeeper)
To get data from kafka I use kafka indexing service (supervisor), which is managed by overlord.
The problem is that recently I started seeing these errors and no data published to dataset:
2018-10-17T04:22:05,547 WARN [task-runner-0-priority-0] io.druid.indexing.kafka.KafkaIndexTask - OffsetOutOfRangeException with message [Offsets out of range with no configured reset policy for partitions: {request-staging-0=4774}] 2018-10-17T04:22:05,547 WARN [task-runner-0-priority-0] io.druid.indexing.kafka.KafkaIndexTask - Retrying in 30000ms 2018-10-17T04:22:35,550 WARN [task-runner-0-priority-0] io.druid.indexing.kafka.KafkaIndexTask - OffsetOutOfRangeException with message [Offsets out of range with no configured reset policy for partitions: {request-staging-0=4774}] 2018-10-17T04:22:35,550 WARN [task-runner-0-priority-0] io.druid.indexing.kafka.KafkaIndexTask - Retrying in 30000ms 2018-10-17T04:23:05,553 WARN [task-runner-0-priority-0] io.druid.indexing.kafka.KafkaIndexTask - OffsetOutOfRangeException with message [Offsets out of range with no configured reset policy for partitions: {request-staging-0=4774}]
After some research I found that I need to set this property in kafka consumer properties
auto.offset.reset=latestAnd also I see in the log file of indexing task that property is set to none:
2018-10-17T03:01:27,769 INFO [task-runner-0-priority-0] org.apache.kafka.clients.consumer.ConsumerConfig - ConsumerConfig values: auto.commit.interval.ms = 5000 auto.offset.reset = none bootstrap.servers = [kafka.myhost.com:9092]
Unfortunately I didn't find how to modify it in kafka, because on kafka server I run only kafka server like this:
/home/centos/kafka/bin/kafka-server-start.sh -daemon /home/centos/devops/kafka/config/server.propertiesand I don't know where to specify consumer properties. Please let me know if you know.
But I found another way how to fix this problem. You can also specify one setting in supervisor config. Here is what you need to set:
{ "type": "kafka", "dataSchema": { ... }, "tuningConfig" : { "type": "kafka", "resetOffsetAutomatically": true }, "ioConfig": { ... } }
and resubmit it to overlord:
curl -XPOST -H'Content-Type: application/json' -d @/fullpath/supervisor-config.json http://localhost:8090/druid/indexer/v1/supervisor
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment