consumer hang fix (Closes #1065) #1066
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
After few days of experiments found the root cause.
### Few details on the fix:
KafkaConsumerManager's `"Consumer Expiration Thread"` acquires a lock on (`KafkaConsumerManager.this` instance) and calls `expired` method of `KafkaConsumerState.java`. The `Consumer Expiration Thread` sometimes cannot get the lock on `expired` is because if there is a consumer consuming records there is a lock on the `KafkaConsumerState instance` which may take longer to release depending on the time taken in the IO to fetch the records from kafka broker.
From the code it seems like the double lock for `expired and updateExpiration` is not required.
### Thread dump snippets
{code}
"qtp1551629761-29" prio=5 Id=29 BLOCKED on io.confluent.kafkarest.v2.KafkaConsumerManager@3a09eed4 owned by "Consumer Expiration Thread" Id=47
at app//io.confluent.kafkarest.v2.KafkaConsumerManager.getConsumerInstance(KafkaConsumerManager.java:669)
- blocked on io.confluent.kafkarest.v2.KafkaConsumerManager@3a09eed4
at app//io.confluent.kafkarest.v2.KafkaConsumerManager.getConsumerInstance(KafkaConsumerManager.java:679)
"qtp1551629761-35" prio=5 Id=35 BLOCKED on io.confluent.kafkarest.v2.KafkaConsumerManager@3a09eed4 owned by "Consumer Expiration Thread" Id=47
at app//io.confluent.kafkarest.v2.KafkaConsumerManager.createConsumer(KafkaConsumerManager.java:180)
- blocked on io.confluent.kafkarest.v2.KafkaConsumerManager@3a09eed4
at app//com.nuance.coretech.fabric.kafka.rest.plugin.resources.ConsumerResources.createGroup(ConsumerResources.java:116)
"Consumer Expiration Thread" daemon prio=5 Id=47 BLOCKED on io.confluent.kafkarest.v2.JsonKafkaConsumerState@4d2bbeda owned by "pool-4-thread-13" Id=153
at app//io.confluent.kafkarest.v2.KafkaConsumerState.expired(KafkaConsumerState.java:345)
- blocked on io.confluent.kafkarest.v2.JsonKafkaConsumerState@4d2bbeda
at app//io.confluent.kafkarest.v2.KafkaConsumerManager$ExpirationThread.run(KafkaConsumerManager.java:761)
- locked io.confluent.kafkarest.v2.KafkaConsumerManager@3a09eed4
"pool-4-thread-13" prio=5 Id=153 RUNNABLE (in native)
at [email protected]/sun.nio.ch.EPoll.wait(Native Method)
at [email protected]/sun.nio.ch.EPollSelectorImpl.doSelect(EPollSelectorImpl.java:120)
at [email protected]/sun.nio.ch.SelectorImpl.lockAndDoSelect(SelectorImpl.java:124)
- locked sun.nio.ch.Util$2@1458190c
- locked sun.nio.ch.EPollSelectorImpl@3ebca1e6
at [email protected]/sun.nio.ch.SelectorImpl.select(SelectorImpl.java:136)
at app//org.apache.kafka.common.network.Selector.select(Selector.java:869)
at app//org.apache.kafka.common.network.Selector.poll(Selector.java:465)
at app//org.apache.kafka.clients.NetworkClient.poll(NetworkClient.java:561)
at app//org.apache.kafka.clients.consumer.internals.ConsumerNetworkClient.poll(ConsumerNetworkClient.java:265)
{code}
|
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
After few days of experiments found the root cause.
Closes #1065
Few details on the fix:
KafkaConsumerManager's
"Consumer Expiration Thread"acquires a lock on (KafkaConsumerManager.thisinstance) and callsexpiredmethod ofKafkaConsumerState.java. TheConsumer Expiration Threadsometimes takes time to get the lock onsynchronized expiredbecause there can be a consumer consuming records that has a lock on theKafkaConsumerState instancewhich may take longer to release depending on the time taken in the IO to fetch the records from kafka broker. The situation can get worse with more number of active consumers fetching records.The two locks for
expired and updateExpirationare not required.Thread dump snippets