Feat/poc continuous profiling #4556

lbloder · 2025-07-15T20:00:54Z

📜 Description

💡 Motivation and Context

Initial implementation of #2635

💚 How did you test it?

📝 Checklist

I added tests to verify the changes.
No new PII added or SDK only sends newly added PII if sendDefaultPII is enabled.
I updated the docs if needed.
I updated the wizard if needed.
Review from the native team if needed.
No breaking change or entry added to the changelog.
No breaking change for hybrid SDKs or communicated to hybrid SDKs.

🔮 Next steps

Investigate why TRACE lifecycle profiles do not show up in Sentry (Support ProfileLifecycle.TRACE #4576)
Add external option for profile lifecycle (Support ProfileLifecycle.TRACE #4576)

…sed on jfr converter bundled with asyncprofiler

… use existing SentryStackFrame instead of JfrFrame,

…inuous profiler in sentry init

…t in SentrySpan to work around scientific notation of double, use wall clock profiling

# Conflicts: # sentry/build.gradle.kts # sentry/src/test/java/io/sentry/ExternalOptionsTest.kt # sentry/src/test/java/io/sentry/JsonSerializerTest.kt # sentry/src/test/java/io/sentry/SentryClientTest.kt # sentry/src/test/java/io/sentry/SentryOptionsTest.kt

…using external options

…d of file extension

…ofilerTest

github-actions · 2025-07-15T20:01:26Z

	Fails
🚫	Please consider adding a changelog entry for the next release.

	Messages
📖	Do not forget to update Sentry-docs with your feature once the pull request gets approved.

Instructions and example for changelog

Please add an entry to CHANGELOG.md to the "Unreleased" section. Make sure the entry includes this PR's number.

Example:

## Unreleased

- Feat/poc continuous profiling ([#4556](https://github.com/getsentry/sentry-java/pull/4556))

If none of the above apply, you can opt out of this check by adding #skip-changelog to the PR description.

Generated by 🚫 dangerJS against 03a20dd

github-actions · 2025-07-15T20:17:42Z

Performance metrics 🚀

	Plain	With Sentry	Diff
Startup time	447.81 ms	457.62 ms	9.80 ms
Size	1.58 MiB	2.09 MiB	520.14 KiB

sentry/src/main/java/io/sentry/protocol/profiling/JfrToSentryProfileConverter.java

sentry/src/main/java/io/sentry/protocol/profiling/JfrSample.java

sentry/src/main/java/io/sentry/protocol/profiling/JfrFrame.java

.../src/main/java/io/sentry/asyncprofiler/provider/AsyncProfilerContinuousProfilerProvider.java

lbloder · 2025-07-22T08:18:25Z

.../src/main/java/io/sentry/asyncprofiler/convert/JfrAsyncProfilerToSentryProfileConverter.java

+              long divisor = jfr.ticksPerSec / 1000_000_000L;
+              long myTimeStamp =
+                  jfr.chunkStartNanos + ((event.time - jfr.chunkStartTicks) / divisor);
+
+              JfrSample sample = new JfrSample();
+              Instant instant = Instant.ofEpochSecond(0, myTimeStamp);
+              double timestampDouble =
+                  instant.getEpochSecond() + instant.getNano() / 1_000_000_000.0;
+
+              sample.timestamp = timestampDouble;


Revisiting this in follow-up PR, I think the way the timestamp is calculated is not correct

Reworked in follow-up (#4576)

adinauer · 2025-07-22T10:07:30Z

@sentry review

.../src/main/java/io/sentry/asyncprofiler/convert/JfrAsyncProfilerToSentryProfileConverter.java

...y-async-profiler/src/main/java/io/sentry/asyncprofiler/profiling/JavaContinuousProfiler.java

.../src/main/java/io/sentry/asyncprofiler/convert/JfrAsyncProfilerToSentryProfileConverter.java

seer-by-sentry · 2025-07-22T10:12:36Z

sentry/src/main/java/io/sentry/SentryEnvelopeItem.java

+
+              if (profileChunk.getPlatform().equals("java")) {
+                final IProfileConverter profileConverter =
+                    ProfilingServiceLoader.loadProfileConverter();
+                if (profileConverter != null) {
+                  try {
+                    final SentryProfile profile =
+                        profileConverter.convertFromFile(traceFile.toPath());
+                    profileChunk.setSentryProfile(profile);
+                  } catch (IOException e) {
+                    throw new SentryEnvelopeException("Profile conversion failed");
+                  }
+                }
+              } else {


The profile conversion logic should have proper error handling and fallback. If conversion fails, the current implementation throws an exception, but it might be better to log the error and continue with a degraded experience.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

seer-by-sentry · 2025-07-22T10:12:39Z

sentry/src/main/java/io/sentry/profiling/ProfilingServiceLoader.java

+      return NoOpContinuousProfiler.getInstance();
+    } catch (Throwable t) {
+      logger.log(
+          SentryLevel.ERROR,
+          "Failed to load continuous profiler provider, using NoOpContinuousProfiler",
+          t);
+      return NoOpContinuousProfiler.getInstance();
+    }


The service loader methods silently catch all Throwable and either return null or a no-op implementation. This could hide important configuration or setup errors. Consider logging the errors at appropriate levels and being more specific about which exceptions to catch.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

seer-by-sentry · 2025-07-22T10:12:41Z

...y-async-profiler/src/main/java/io/sentry/asyncprofiler/profiling/JavaContinuousProfiler.java

+      startProfileChunkTimestamp = new SentryNanotimeDate();
+    }
+    filename = profilingTracesDirPath + File.separator + SentryUUID.generateSentryId() + ".jfr";
+    String startData = null;
+    try {
+      final String profilingIntervalMicros =
+          String.format("%dus", (int) SECONDS.toMicros(1) / profilingTracesHz);
+      final String command =
+          String.format(
+              "start,jfr,event=wall,interval=%s,file=%s", profilingIntervalMicros, filename);
+      System.out.println(command);
+      startData = profiler.execute(command);
+    } catch (Exception e) {
+      logger.log(SentryLevel.ERROR, "Failed to start profiling: ", e);
+    }
+    // check if profiling started
+    if (startData == null) {
+      return;
+    }
+
+    isRunning = true;


The profiler initialization and AsyncProfiler command construction should have more robust error handling. The current implementation may not handle all edge cases properly.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

sentry-samples/sentry-samples-spring-boot-jakarta/src/main/resources/application.properties

sentry/src/main/java/io/sentry/protocol/profiling/JfrToSentryProfileConverter.java

.../src/main/java/io/sentry/asyncprofiler/convert/JfrAsyncProfilerToSentryProfileConverter.java

...y-async-profiler/src/main/java/io/sentry/asyncprofiler/profiling/JavaContinuousProfiler.java

adinauer · 2025-07-24T09:07:00Z

sentry/src/main/java/io/sentry/SentryOptions.java

@@ -578,6 +578,8 @@ public class SentryOptions {

  private @NotNull ISocketTagger socketTagger = NoOpSocketTagger.getInstance();

+  private @Nullable String profilingTracesDirPath;


m Can we simply default this to use File.createTempFile or similar so all that's needed is the sample rate for profiling to work?

Added in follow-up (#4576)

adinauer · 2025-07-24T09:09:32Z

sentry-samples/sentry-samples-spring-boot-jakarta/src/main/resources/application.properties

 sentry.enablePrettySerializationOutput=false
 in-app-includes="io.sentry.samples"
 sentry.logs.enabled=true
+sentry.profile-session-sample-rate=1.0


Doesn't this also need the path set?

Dependency on async profiler also seems to be missing from gradle build file.

Fixed in follow-up (#4576)

sentry/src/main/java/io/sentry/protocol/profiling/SentryProfile.java

sentry/src/main/java/io/sentry/protocol/profiling/ThreadMetadata.java

lcian

Could you also document the commit SHA of async-profiler we're vendoring?
Perhaps as a README.md in sentry-async-profiler.
Might come in handy for troubleshooting/upgrades in the future and could be hard to find the exact commit later.

We could also add the vendored directory path to the ignored ones in our codecov.yml so that it won't impact coverage stats.

lcian · 2025-07-25T15:05:03Z

sentry/src/main/java/io/sentry/SentryEnvelopeItem.java

-              if (base64Trace.isEmpty()) {
-                throw new SentryEnvelopeException("Profiling trace file is empty");
+
+              if (profileChunk.getPlatform().equals("java")) {


Or we can make it into an enum instead of string

sentry/src/main/java/io/sentry/profiling/JavaContinuousProfilerProvider.java

sentry/src/main/java/io/sentry/protocol/profiling/JfrSample.java

lcian · 2025-07-25T15:39:31Z

sentry/src/main/java/io/sentry/protocol/profiling/JfrSample.java

+
+  public int stackId;
+
+  public @Nullable String threadId;


This is documented as required in the dev docs, perhaps we can provide a default value if it's missing, even though I'm not sure if that makes much sense

sentry/src/main/java/io/sentry/protocol/profiling/ThreadMetadata.java

lbloder added 21 commits April 25, 2025 12:14

wip

f32782c

Merge branch 'main' into feat/poc-continuous-profiling

4fd36d2

re-add data category profile_chunk, fix json naming, new converter ba…

0e2a1b7

…sed on jfr converter bundled with asyncprofiler

read java thread ids from jfr and use those instead of os thread ids,…

558ff79

… use existing SentryStackFrame instead of JfrFrame,

adhere to sentry conventions re format null safety etc, fix compilation

04c82de

add profile-session-sample-rate to external options

f44e7fd

add platform as constructor param to ProfileChunk, wip: set java cont…

8b4c71a

…inuous profiler in sentry init

add doubleToBigDecimal in JfrSample and ProfileChunk, same as we do i…

a44287b

…t in SentrySpan to work around scientific notation of double, use wall clock profiling

rename JfrProfile to SentryProfile

9cf7ef6

move java profiling into its own module, load using SPI

5534204

[WIP] use getProfilingTracesDirPath

cabc384

add missing build.gradle.kts for profiler module

ec59901

Merge branch 'main' into feat/poc-continuous-profiling

54c3a8f

WIP continuous profiling in trace mode

543fc40

cleanup unused classes from vendor, cleanup packages

fba1881

allow setting profiling-traces-dir-path independently from cache dir …

55df61c

…using external options

use profileChunk.platform to decide how to deal with the chunk instea…

9bcd4ea

…d of file extension

port relevant AndroidContinuousProfilerTest tests to JavaContinuousPr…

3f83146

…ofilerTest

add service loader tests for profiler and profile converter

4a7403a

remove old jfr test files

03a20dd