continuous-async-profiler

This is a spring boot library that runs async-profiler in the continuous mode.

This project is just a set of tool that run async-profiler.

If you want to use the other version of profiler than 2.9 provided by AP-Loader, you need to specify path to libasyncProfiler.so. You can do it with spring application properties/yaml/... For example:

async-profiler.continuous.profiler-lib-path=/path/to/libasyncProfiler.so

How to add use it in spring boot application?

You just need to add dependency to your spring boot application

<dependency>
    <groupId>com.github.krzysztofslusarski</groupId>
    <artifactId>continuous-async-profiler-spring-starter</artifactId>
    <version>2.1</version>
</dependency>

with a proper version.

How to add use it in spring application without spring boot?

Step 1 - add dependency

<dependency>
    <groupId>com.github.krzysztofslusarski</groupId>
    <artifactId>continuous-async-profiler</artifactId>
    <version>2.1</version>
</dependency>

Step 2 - import configuration

Add a following import in your @Configuration file:

@Import(ContinuousAsyncProfilerConfiguration.class)

How it works on defaults

The async-profiler is run all the time in wall-clock mode. Output from the profiler is dumped to the logs/continuous directory every 60 seconds. The files in the logs/continuous directory are stored for 24h. Once a day files from that directory that matches regex .*_13:0.* are copied to logs/archive directory. On defaults the archive contains 10 minutes of each profiled day. Files in the logs/archive directory are stored for 30 days.

Configuration properties and defaults

Properties manageable at runtime:

• async-profiler.continuous.enabled = true - if the tool should work or not
• async-profiler.continuous.continuous-outputs-max-age-hours = 24h - time in hours, how long to keep files in the continuous directory
• async-profiler.continuous.archive-outputs-max-age-days = 30d - time in days, how long to keep files in the archive directory
• async-profiler.continuous.archive-copy-regex = .*_13:0.* - regex for file name, which files should be copied from the continuous to the archive directory
• async-profiler.continuous.event = wall - async-profiler event to fetch
• async-profiler.continuous.stop-work-file = profiler-stop - path to a file, if the file exists then profiler is not running, using this file you can turn on/off profiling at runtime

Properties not manageable at runtime:

• async-profiler.continuous.load-native-library = true - if the tool should load native async-profiler library (turning off disables starter permanently)
• async-profiler.continuous.dump-interval = 60s - time in seconds, how often tool should dump profiler outputs
• async-profiler.continuous.output-dir.continuous = logs/continuous - where continuous output should be stored
• async-profiler.continuous.output-dir.archive = logs/archive - where archive of the outputs should be stored
• async-profiler.continuous.profiler-lib-path - path to libasyncProfiler.so
• async-profiler.continuous.manageable-properties-repository = default - what properties resources should be used
  • default - properties from spring context
  • jmx - properties from spring context with registered mbean for changing them in runtime

Changing properties source

You can change manageable properties source from spring to any you want. You need to implement a spring bean, that implements following interface:

com.github.krzysztofslusarski.asyncprofiler.ContinuousAsyncProfilerManageablePropertiesRepository

Troubleshooting

First of all this tool is just an async-profiler runner. If you cannot run plain async-profiler on your OS then you cannot use this tool.