Skip to main content

Java

Pyroscope uses async-profiler to collect profiling data from Java applications.

Supported platforms#

LinuxmacOSWindowsDocker

Installation#

Java integration is distributed as a single jar file: pyroscope.jar. It contains native async-profiler libraries for:

  • Linux on x64;
  • Linux on ARM64;
  • MacOS on x64.
  • MacOS on ARM64.

Visit our GitHub releases page to download the latest version of pyroscope.jar.

The latest release is also available on Maven Central.

Profiling Java application#

You can start pyroscope either from your apps's java code or attach it as javaagent

Start pyroscope from app's java code#

Add pyroscope dependency:

<dependency>  <groupId>io.pyroscope</groupId>  <artifactId>agent</artifactId>  <version>pyroscope_version</version></dependency>

Then add the following code to your application:

PyroscopeAgent.start(    new Config.Builder()      .setApplicationName("ride-sharing-app-java")      .setProfilingEvent(EventType.ITIMER)      .setFormat(Format.JFR)      .setServerAddress("http://pyroscope-server:4040")      // Optionally, if authentication is enabled, specify the API key.      // .setAuthToken(System.getenv("PYROSCOPE_AUTH_TOKEN"))      .build()  );

You can also optionally replace some pyroscope components

PyroscopeAgent.start(  new PyroscopeAgent.Options.Builder(config)    .setExporter(snapshot -> {      // Your custom export/upload logic may go here      // It is invoked every 10 seconds by default with snapshot of       // profiling data    })    .setLogger((l, msg, args) -> {      // Your custom logging may go here      // Pyroscope does not depend on any logging library      System.out.printf((msg) + "%n", args);    })    .setScheduler(profiler -> {      // Your custom profiling schedule logic may go here    })    .build());

Start pyroscope as javaagent#

To start profiling a Java application, run your application with pyroscope.jar javaagent:

export PYROSCOPE_APPLICATION_NAME=my.java.appexport PYROSCOPE_SERVER_ADDRESS=http://pyroscope-server:4040
# Optionally, if authentication is enabled, specify the API key.# export PYROSCOPE_AUTH_TOKEN={YOUR_API_KEY}
java -javaagent:pyroscope.jar -jar app.jar

Configuration#

When you start pyroscope as javaagent or obtain configuration by Config.build() pyroscope searches for configuration in multiple sources: system properties, environment variables, pyroscope.properties file. Properties keys has same name as environment variables, but lowercased and replaced _ with ., so PYROSCOPE_FORMAT becomes pyroscope.format

Java integration supports JFR format to be able to support multiple events (JFR is the only output format that supports multiple events in async-profiler). There are several environment variables that define how multiple event configuration works:

  • PYROSCOPE_FORMAT sets the profiler output format. The default is collapsed, but in order to support multiple formats it must be set to jfr.
  • PYROSCOPE_PROFILER_EVENT sets the profiler event. With JFR format enabled, this event refers to one of the possible CPU profiling events: itimer, cpu, wall. The default is itimer.
  • PYROSCOPE_PROFILER_ALLOC sets the allocation threshold to register the events, in bytes (equivalent to --alloc= in async-profiler). The default value is "" - empty string, which means that allocation profiling is disabled. Setting it to 0 will register all the events.
  • PYROSCOPE_PROFILER_LOCK sets the lock threshold to register the events, in nanoseconds (equivalent to --lock= in async-profiler). The default value is "" - empty string, which means that lock profiling is disabled. Setting it to 0 will register all the events.

There are some extra configuration options:

  • PYROSCOPE_CONFIGURATION_FILE sets an additional properties configuration file. The default value is pyroscope.properties.
  • PYROSCOPE_BASIC_AUTH_USER HTTP Basic authentication username. The default value is "" - empty string, no authentication.
  • PYROSCOPE_BASIC_AUTH_PASSWORD HTTP Basic authentication password. The default value is "" - empty string, no authentication.
  • PYROSCOPE_TENANT_ID phlare tenant ID, passed as X-Scope-OrgID http header. The default value is "" - empty string, no tenant ID.
  • PYROSCOPE_HTTP_HEADERS extra http headers in json format, for example: {"X-Header": "Value"}. The default value is {} - no extra headers.
  • PYROSCOPE_LABELS sets static labels in the form of comma separated key=value pairs. The default value is "" - empty string, no labels.
  • PYROSCOPE_LOG_LEVEL determines the level of verbosity for Pyroscope's logger. Available options include debug, info, warn, and error. The default value is set to info.
  • PYROSCOPE_PUSH_QUEUE_CAPACITY specifies the size of the ingestion queue that temporarily stores profiling data in memory during network outages. It is recommended to keep this queue size reasonably small, and the default value is set to 8.
  • PYROSCOPE_INGEST_MAX_TRIES sets the maximum number of times to retry an ingestion API call in the event of failure. A value of -1 indicates that the retries will continue indefinitely. The default value is set to 8.
  • PYROSCOPE_EXPORT_COMPRESSION_LEVEL_JFR sets the level of GZIP compression applied to uploaded JFR files. This option accepts values of NO_COMPRESSION, BEST_SPEED, BEST_COMPRESSION, and DEFAULT_COMPRESSION, as well as numeric values between -1 and 9, where -1 represents DEFAULT_COMPRESSION, 0 represents NO_COMPRESSION, 1 represents BEST_SPEED, and 9 represents BEST_COMPRESSION. The default value is set to BEST_SPEED.
  • PYROSCOPE_EXPORT_COMPRESSION_LEVEL_LABELS perates similarly to PYROSCOPE_EXPORT_COMPRESSION_LEVEL_JFR, but applies to the dynamic labels part. The default value is set to BEST_SPEED.
  • PYROSCOPE_ALLOC_LIVE is a boolean value that enables live object profiling when set to true. It is disabled by default.
  • PYROSCOPE_GC_BEFORE_DUMP is a boolean value that executes a System.gc() command before dumping the profile when set to true. This option may be useful for live profiling, but is disabled by default.

Labels#

It is possible to add dynamic tags (labels) to the profiling data. These tags can be used to filter the data in the UI.

Add labels dynamically:

Pyroscope.LabelsWrapper.run(new LabelsSet("controller", "slow_controller"), () -> {  slowCode();});

It is also possible to possible to add static tags (labels) to the profiling data:

Pyroscope.setStaticLabels(Map.of("region", System.getenv("REGION")));// or with Config.Builder if you start pyroscope with PyroscopeAgent.startPyroscopeAgent.start(new Config.Builder()    .setLabels(mapOf("region", System.getenv("REGION")))    // ...    .build());

Sending data to Phlare with Pyroscope java integration#

Starting with weekly-f8 you can ingest pyroscope profiles directly to phlare.

PyroscopeAgent.start(  new Config.Builder()    .setApplicationName("phlare.java.app")    .setFormat(Format.JFR)    .setServerAddress("<URL>")    // Optional HTTP Basic authentication    .setBasicAuthUser("<User>")    .setBasicAuthPassword("<Password>")    // Optional Phlare tenant ID    .setTenantID("<TenantID>")    .build());

To configure java integration to send data to Phlare, replace the <URL> placeholder with the appropriate server URL. This could be the grafana.com Phlare URL or your own custom Phlare server URL.

If you need to send data to grafana.com, you'll have to configure HTTP Basic authentication. Replace <User> with your grafana.com stack user and <Password> with your grafana.com API key.

If your Phlare server has multi-tenancy enabled, you'll need to configure a tenant ID. Replace <TenantID> with your Phlare tenant ID.

Java profiling examples#

Check out the following resources to learn more about Java profiling:

Frame width represents CPU time per function
Pyroscope