Spring boot lambda starts twice with dd-trace-java library

[pacome2018]

Hello,

I'm having troubles with Spring Boot in Lambda when adding the Datadog tracer to my function.

As we can see in the log file, Spring Boot restarts after a successful but void first start.

I don't see anything suspisious between the two boots, maybe someone can help with the logs ?

Also, it used to work fine, problem appeared recently.
(I redeployed my AWS stacks. And now I can reproduce this issue with a new stack always with no changes on our side)

I tried with following versions :
dd-trace-java:8 (= v1.5.0) & Datadog-Extension-ARM:36 (and 37)
dd-trace-java:7 (= v1.1.3) & Datadog-Extension-ARM:34
tracer versions according to this project https://github.com/DataDog/datadog-lambda-java

log-spring-lambda-datadog.csv
Datadog tracer debugger is enabled.

Interesting line numbers for the logs :
l. 1255 Spring Boot 1st start
l. ~1310 Spring Boot stop
l. 2647 Spring Boot 2nd start

Interesting logs :
[dd-task-scheduler] WARN datadog.telemetry.dependency.DependencyResolverQueue - unable to detect dependency for URI file:/var/task/
[dd-task-scheduler] DEBUG datadog.telemetry.dependency.DependencyResolverQueue - unable to detect dependency for URI jrt:/jdk.localedata

[devinsba]

Thanks for the report, someone will take a look. In case those errors are the cause of the issue to you can work around these errors by disabling that feature:

Environment variable: DD_INSTRUMENTATION_TELEMETRY_ENABLED=false

System property: dd.instrumentation.telemetry.enabled=false

[pacome2018]

Thanks for the response, DD_INSTRUMENTATION_TELEMETRY_ENABLED=false did clear the warnings but the error remains.
Enabling Spring traces doesn't give interesting information.

[pacome2018]

We found out the problem was the Lambda had not enough memory to start SpringBoot after adding dd-tracer layer.

Sorry for the trouble, hope it helps if someone stumble upons this 🙂