Profiling in Docker container with direct connection

This article describes how to set up profiling of Java application running in a Docker container.

The container will be configured in such a way, that local port *:10001 is mapped to the port of the profiler agent running inside the Docker container. You will be able to connect profiler UI to this port using direct discovery method.

On Docker side

1. Add few lines to your Dockerfile:

• Install YourKit Java Profiler agents:

  RUN wget https://www.yourkit.com/download/docker/YourKit-JavaProfiler-2022.9-docker.zip -P /tmp/ && \
    unzip /tmp/YourKit-JavaProfiler-2022.9-docker.zip -d /usr/local && \
    rm /tmp/YourKit-JavaProfiler-2022.9-docker.zip

• Load the agent to the JVM by adding the Java command line option -agentpath.

  For example, if you start your application with

  java -jar my-app.jar

  ...do it like this:

  java -agentpath:/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-x86-64/libyjpagent.so=port=10001,listen=all -jar my-app.jar

  In this example:

  • The agent is configured to listen port 10001. The same port should be used in the -p option when running the container (see below). If you decide to use a different port, please ensure you have changed it everywhere.

  • Profiler agent is a Linux 64-bit agent, compiled for glibc. If you run Docker on another platform, please choose appropriate agent path from the table below:

    Platform		Agent path
    Linux (glibc)	x86, 64-bit	/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-x86-64/libyjpagent.so
    	ARM 64-bit (AArch64)	/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-arm-64/libyjpagent.so
    	ppc64le, 64-bit, little-endian	/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-ppc-64le/libyjpagent.so
    Alpine Linux (musl)	x86, 64-bit	/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-musl-x86-64/libyjpagent.so
    	ARM 64-bit (AArch64)	/usr/local/YourKit-JavaProfiler-2022.9/bin/linux-musl-arm-64/libyjpagent.so
    Windows	x86, 64-bit	<profiler directory>\bin\windows-x86-64\yjpagent.dll
    	ARM 64-bit (AArch64)	<profiler directory>\bin\windows-arm-64\yjpagent.dll

    Please find detailed description of how to specify -agentpath and choose the agent startup options here.

2. Build the container for the modified Dockerfile.

3. When running the container, make the agent port visible with the option -p:

docker run -p 10001:10001 my_container

Connect to the profiled application

When the application is running in the container, connect to it from the profiler UI using the direct discovery method. You will need to specify IP address of your Docker container and 10001 as a port number. If you are running Docker locally on your developer machine, please use localhost as a container address.

Notes on a special case: a container with restricted home directory

For security reasons, the user home directory in a docker container may not exist or be read-only.

In this case, the agent will use /tmp instead of <user home>/Snapshots as the default snapshot directory. The snapshot directory may be overridden with the startup option dir.

The profiler agent log file neither can be created in its normal location <user home>/.yjp/log. Should you need the agent log file for troubleshooting purposes, specify a writable directory for it with the startup option logdir.