Analyzing ConsoleAppender Blocking and Using AsyncLogger for High-Performance Logging in Log4j2

Author: Du Yunjie, architect, responsible for service governance, MQ, cloud platform, APM, IM, distributed tracing, monitoring, configuration center, distributed task scheduling, ID generation, and distributed locks.

1. Background

Several services experienced deadlock situations, which were traced to the use of the ConsoleAppender of the logging framework in production.

2. Investigation Process

2.1 Stack of Blocked Threads

Thread dumps obtained via jstack showed many threads blocked on the monitor lock of

org.apache.logging.log4j.core.layout.TextEncoderHelper.copyDataToDestination

. An example stack trace is shown below:

The block occurs when a thread attempts to acquire the lock of an OutputStreamManager (a subclass of ByteBufferDestination).

2.2 Stack of Threads Holding the Lock

The thread that holds the lock was also captured, revealing that business threads using ConsoleAppender need to synchronize on the OutputStreamManager monitor, then invoke PrintStream.write() → BufferedOutputStream.write() → FileOutputStream.write() and finally flush(). Only after the console write succeeds is the lock released, causing other threads to block.

The call sequence is illustrated below:

2.3 Console Log Performance Analysis

The synchronization around the console write is inefficient; benchmark data from the official site shows very low throughput for console logging.

3. AsyncLogger

Using a synchronous logger with ConsoleAppender blocks business threads, while the same applies to RollingRandomAccessFileAppender but with higher efficiency. The solution is to switch to AsyncLogger, which leverages the high‑performance Disruptor ring‑buffer to make logging asynchronous, eliminating lock contention and dramatically improving response time.

AsyncLogger workflow: business threads publish log events to the Disruptor ring buffer via tryPublishEvent() and return immediately without acquiring locks. A dedicated thread pool consumes the events and writes them to the destination.

Performance figures from the official documentation demonstrate the superiority of AsyncLogger.

3.1 How to Integrate

1) Add the Disruptor dependency:

<dependency>
  <groupId>com.lmax</groupId>
  <artifactId>disruptor</artifactId>
  <version>3.4.2</version>
</dependency>

2) Configure Log4j2 XML as follows:

<Loggers>
  <!-- Set Spring logs to INFO -->
  <AsyncLogger name="org.springframework" level="ERROR" includeLocation="true"/>
  <asyncRoot level="INFO" includeLocation="true">
    <AppenderRef ref="INFO"/>
    <AppenderRef ref="WARN"/>
    <AppenderRef ref="ERROR"/>
  </asyncRoot>
</Loggers>

4. Conclusion

Do not use ConsoleAppender in production.

Strongly recommend AsyncLogger for high‑performance logging.

Synchronous loggers can be used, but RollingRandomAccessFileAppender is preferred over ConsoleAppender due to higher efficiency.

The company has enforced a ban on ConsoleAppender and mandates AsyncLogger, eliminating service deadlocks caused by logging.