amuthan Ganeshan created KAFKA-9073:
---------------------------------------
Summary: Kafka Streams State stuck in rebalancing after one of the
StreamThread encounters IllegalStateException
Key: KAFKA-9073
URL: https://issues.apache.org/jira/browse/KAFKA-9073
Project: Kafka
Issue Type: Bug
Components: streams
Affects Versions: 2.3.0
Reporter: amuthan Ganeshan
I have a Kafka stream application that stores the incoming messages into a
state store, and later during the punctuation period, we store them into a big
data persistent store after processing the messages.
The application consumes from 120 partitions distributed across 40 instances.
The application has been running fine without any problem for months, but all
of a sudden some of the instances failed because of a stream thread exception
saying
```java.lang.IllegalStateException: No current assignment for partition
<app_name>-<store_name>-changelog-98```
And other instances are stuck in the REBALANCING state, and never comes out of
it. Here is the full stack trace, I just masked the application-specific app
name and store name in the stack trace due to NDA.
```
2019-10-21 13:27:13,481 ERROR
[application.id-a2c06c51-bfc7-449a-a094-d8b770caee92-StreamThread-3]
[org.apache.kafka.streams.processor.internals.StreamThread] [] stream-thread
[application.id-a2c06c51-bfc7-449a-a094-d8b770caee92-StreamThread-3]
Encountered the following error during processing:
java.lang.IllegalStateException: No current assignment for partition
application.id-store_name-changelog-98
at
org.apache.kafka.clients.consumer.internals.SubscriptionState.assignedState(SubscriptionState.java:319)
at
org.apache.kafka.clients.consumer.internals.SubscriptionState.requestFailed(SubscriptionState.java:618)
at
org.apache.kafka.clients.consumer.internals.Fetcher$2.onFailure(Fetcher.java:709)
at
org.apache.kafka.clients.consumer.internals.RequestFuture.fireFailure(RequestFuture.java:177)
at
org.apache.kafka.clients.consumer.internals.RequestFuture.raise(RequestFuture.java:147)
at
org.apache.kafka.clients.consumer.internals.RequestFutureAdapter.onFailure(RequestFutureAdapter.java:30)
at
org.apache.kafka.clients.consumer.internals.RequestFuture$1.onFailure(RequestFuture.java:209)
at
org.apache.kafka.clients.consumer.internals.RequestFuture.fireFailure(RequestFuture.java:177)
at
org.apache.kafka.clients.consumer.internals.RequestFuture.raise(RequestFuture.java:147)
at
org.apache.kafka.clients.consumer.internals.ConsumerNetworkClient$RequestFutureCompletionHandler.fireCompletion(ConsumerNetworkClient.java:574)
at
org.apache.kafka.clients.consumer.internals.ConsumerNetworkClient.firePendingCompletedRequests(ConsumerNetworkClient.java:388)
at
org.apache.kafka.clients.consumer.internals.ConsumerNetworkClient.poll(ConsumerNetworkClient.java:294)
at
org.apache.kafka.clients.consumer.internals.ConsumerNetworkClient.poll(ConsumerNetworkClient.java:233)
at
org.apache.kafka.clients.consumer.KafkaConsumer.pollForFetches(KafkaConsumer.java:1281)
at
org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1225)
at
org.apache.kafka.clients.consumer.KafkaConsumer.poll(KafkaConsumer.java:1201)
at
org.apache.kafka.streams.processor.internals.StreamThread.maybeUpdateStandbyTasks(StreamThread.java:1126)
at
org.apache.kafka.streams.processor.internals.StreamThread.runOnce(StreamThread.java:923)
at
org.apache.kafka.streams.processor.internals.StreamThread.runLoop(StreamThread.java:805)
at
org.apache.kafka.streams.processor.internals.StreamThread.run(StreamThread.java:774)
```
Now I checked the state sore disk usage; it is less than 40% of the total disk
space available. Restarting the application solves the problem for a short
amount of time, but the error popping up randomly on some other instances
quickly. I tried to change the retry and retry.backoff.ms configuration but not
helpful at all
```
retries = 2147483647
retry.backoff.ms
```
After googling for some time I found there was a similar bug reported to the
Kafka team in the past, and also notice my stack trace is exactly matching with
the stack trace of the reported bug.
Here is the link for the bug reported on a comparable basis a year ago.
https://issues.apache.org/jira/browse/KAFKA-7181
Now I am wondering is there a workaround for this bug though configuration
changes, or is there something wrong the way I set up the application, the
following are the configuration I have for my stream application.
```
consumer.session.timeout.ms=30000
metric.reporters=org.apache.kafka.common.metrics.JmxReporter
replication.factor=3
metadata.max.age.ms=30000
max.partition.fetch.bytes=2000000
producer.retries=2147483647
bootstrap.servers= <bootstrap server list goes here>
metrics.recording.level=DEBUG
producer.retry.backoff.ms=60000
consumer.auto.offset.reset=latest
application.server=0.0.0.0:6063
num.standby.replicas=1
max.poll.records=2
group.initial.rebalance.delay.ms=30000
state.dir= <state dir path goes here>
heartbeat.interval.ms=10000
max.poll.interval.ms=300000
num.stream.threads=10
application.id= <application id goes here>
```
Note: The original bug reported a year back got a conclusion that it is related
to https://issues.apache.org/jira/browse/KAFKA-7657 and reported solved in
version 2.2.0, but I am using the latest 2.3.0 version.
I appreciate your help concerning this bug.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
