问题
I'm running some sample code from http://www.javaworld.com/article/3060078/big-data/big-data-messaging-with-kafka-part-1.html?page=2, and the kafkaconsumer is consuming from topic as desired, but every poll results in print (to std out) of many debug logs, which I don't want.
I have tried changing all INFO and DEBUG to ERROR (even did a grep to make sure) in /config/log4j.properties
, in particular setting log4j.logger.kafka=ERROR
, kafkaAppender, but the problem persists. I referred to How to configure logging for Kafka producers?, and adopted the solution there, but perhaps the situation is different for consumers?
The DEBUG messages all have a similar format:
[Thread-0] DEBUG org.apache.kafka.clients.consumer.internals.Fetcher - Sending fetch for partitions... to broker... (id: 0 rack: null)
and are appearing at rate of 10 every second or so (changing poll argument to 1000 or even 10000 doesn't help, I tried)
Would really appreciate any help from any expert. Thanks in advance!
Edit: Not sure if it matters, but I added BasicConfigurator.configure();
to my main method, to resolve some other error occurring previously that stopped the Consumer from even starting.
回答1:
Just modify the logging level of the chatty class (chatty interaction).
Since in your logs you see log entries originating from org.apache.kafka.clients.consumer.internals.Fetcher
you can simply adjust the logging level for that logger by adding following line to log4j.properties
:
log4j.logger.org.apache.kafka.clients.consumer.internals.Fetcher=WARN
... or any wider catching logger since these are name spaced:
# adjusting logging for entire Kafka
log4j.logger.org.apache.kafka=WARN
Hope this helps
回答2:
create new config xml file
src/main/resources/logback.xml
<configuration>
<appender name="STDOUT" class="ch.qos.logback.core.ConsoleAppender">
<encoder>
<pattern>%d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n</pattern>
</encoder>
</appender>
<logger name="org.apache.kafka" level="WARN"/>
<logger name="org.apache.kafka.common.metrics" level="WARN"/>
<root level="warn">
<appender-ref ref="STDOUT" />
</root>
</configuration>
回答3:
Not sure if you are talking about kafka-console-consumer commands, if yes this is what I did:
[training@confluent-training-vm ~]$ cd /etc/kafka
[training@confluent-training-vm kafka]$ grep DEBUG *.properties
log4j.properties:# Change to DEBUG or TRACE to enable request logging
log4j.properties:# Access denials are logged at INFO level, change to DEBUG to also
log allowed accesses
tools-log4j.properties:log4j.rootLogger=DEBUG, stderr
So, you just need to edit /etc/kafka/tools-log4j.properties
file and remove DEBUG
(or replace it for exmaple by INFO
and WARM
log levels on above line
tools-log4j.properties:log4j.rootLogger=INFO, WARM, stderr
回答4:
I found the solution under another question which deals with Kafka producers, but it's essentially the same problem: https://stackoverflow.com/a/49532152/2380553
So I just run the following 3 lines at the beginning of my program:
org.apache.log4j.Logger.getLogger("org").setLevel(Level.WARN);
org.apache.log4j.Logger.getLogger("akka").setLevel(Level.WARN);
org.apache.log4j.Logger.getLogger("kafka").setLevel(Level.WARN);
来源:https://stackoverflow.com/questions/44129303/kafka-consumer-outputs-excessive-debug-statements-to-console-ecilpse