Workflow failed: GraalVM native tests (#254)

[github-actions]

See GraalVM native tests #254.

[laurit]

@jeanbisutti could you take a look. Could be because of #10607

[jeanbisutti]

It seems relatef to #10457

[jeanbisutti]

The problem is related to #10457 using a SnakeYAML feature not working with GraalVM. The issue may be fixed with #10636 replacing the JAR-based resource extraction by a manifest-based resource extraction:

opentelemetry-java-instrumentation/instrumentation/spring/spring-boot-autoconfigure/src/main/java/io/opentelemetry/instrumentation/spring/autoconfigure/resources/OtelResourceAutoConfiguration.java

Line 54 in 63d734d

return new ManifestResourceProvider();

[jeanbisutti]

The problem is related to #10457 using a SnakeYAML feature not working with GraalVM. The issue may be fixed with #10636 replacing the JAR-based resource extraction by a manifest-based resource extraction:

opentelemetry-java-instrumentation/instrumentation/spring/spring-boot-autoconfigure/src/main/java/io/opentelemetry/instrumentation/spring/autoconfigure/resources/OtelResourceAutoConfiguration.java

Line 54 in 63d734d

return new ManifestResourceProvider();

Or it may be fixed from #10621. @zeitlinger From which PR the manifest-based resource detection replaced the JAR-based one?

[zeitlinger]

It will be fixed in #10636

[github-actions]

See GraalVM native tests #255.

[jeanbisutti]

I have just pushed on #10636. The GraalVM native tests have run successfully on my machine with this PR. @laurit @trask Could you please review this PR? So, the daily GraalVM native tests could be green again.

[trask]

I have just pushed on #10636

is there reason not to send a separate PR to fix? generally faster for us to merge a standalone fix

[zeitlinger]

I have just pushed on #10636

is there reason not to send a separate PR to fix? generally faster for us to merge a standalone fix

I've done that now: #10685

[jeanbisutti]

The problem is related to #10457 using a SnakeYAML feature not working with GraalVM. The issue may be fixed with #10636 replacing the JAR-based resource extraction by a manifest-based resource extraction:

opentelemetry-java-instrumentation/instrumentation/spring/spring-boot-autoconfigure/src/main/java/io/opentelemetry/instrumentation/spring/autoconfigure/resources/OtelResourceAutoConfiguration.java

Line 54 in 63d734d

return new ManifestResourceProvider();

As mentioned before, a problem is related to the SnakeYAML usage with GraalVM native. SnakeYAML is removed from the OpenTelemetry starter in #10636.

There is a second problem causing the failure of the OpenTelemetry starter native tests (introduced in #10602). The second problem can be fixed with 2d7254f:

@DisabledInNativeImage // Without this the native tests if the OtelSpringStarterSmokeTest class will
// fail with org.h2.jdbc.JdbcSQLSyntaxErrorException: Table "TEST_TABLE"
// already exists
class OtelSpringStarterDisabledSmokeTest {

[jeanbisutti]

The GraalVM native tests are passing again on the main branch: https://github.com/open-telemetry/opentelemetry-java-instrumentation/actions/workflows/native-tests-daily.yml