[jira] [Created] (KAFKA-10065) One of the Kafka broker processes in the cluster suddenly hangs and becomes unresponsive

["akihiro kumabe (Jira)" <ji...@apache.org>]

akihiro kumabe created KAFKA-10065:
--------------------------------------

             Summary: One of the Kafka broker processes in the cluster suddenly hangs and becomes unresponsive
                 Key: KAFKA-10065
                 URL: https://issues.apache.org/jira/browse/KAFKA-10065
             Project: Kafka
          Issue Type: Bug
    Affects Versions: 1.1.1
            Reporter: akihiro kumabe


Kafka version: 1.1.1


I have 3 Brokers and 3 zookeepers.
One of the Kafka broker processes in the cluster hung and became unresponsive three times in the last month.
I investigated a Java thread dump. Although some threads such as Kafka Fetcher were Parked, I could not find any deadlock.
Server resources, heap memory seemed to be fine, and Kafka logs also had no suspicious output.
Many TCP connections with close-wait status remained.
Recovered by simply restarting the process.

Similar to # 5778, but it's unclear if a version upgrade will solve it
https://issues.apache.org/jira/browse/KAFKA-5778



--
This message was sent by Atlassian Jira
(v8.3.4#803005)