Docs » Set up Splunk APM » Troubleshoot your instrumentation

Troubleshoot your instrumentation 🔗

If you have instrumented an application but are not seeing the relevant data in Splunk APM, use the following guidelines to troubleshoot your instrumentation.

General troubleshooting 🔗

  1. If you aren’t seeing your instrumented APM service endpoints, verify that the span.kind tag is being set correctly. For instance, client spans should have span.kind = CLIENT and server spans should have span.kind = SERVER. To learn more about span.kind, see the tracing API specification on GitHub:

  2. Make sure the application for the relevant runtime is active. Traces can only be captured from the inbound traffic to the application.

  3. Check that the runtime to which you attached the instrumentation agent has started. For example, for Java you can run jps -lvm in the command line to list all active Java processes. The output of the command is the list of the JVMs currently running in the machine. Make sure the JVM in question appears among them. If not, the JVM failed to start, and you should check the JVM or application logs to find the cause of the problem.

  4. Check that the instrumentation is applied to the runtime. For example, for Java you can confirm that the javaagent was correctly attached in the configuration by exposing the following:

    me@my-application:~$ jps -lvm
    24910 root.war -Xmx512m -ea
    24914 jdk.jcmd/ -lvm
    -Dapplication.home=/usr/lib/jvm/jdk-11.0.10 -Xms8m
    my-application:~ me$

    If the output above doesn’t contain the -javaagent section for the JVM in question, the parameter specified in the startup script was not picked up. To proceed, debug the startup scripts to see where the added or modified parameters are being lost.

  5. Check that the additional startup parameters are correctly passed to the runtime. When the parameters are passed correctly, the service name appears in the value of For instance, the example above refers to,environment=test.

If steps 1-5 don’t help you with your problem, enable debugging mode for the instrumentation, then check the STDOUT and STDERR logs. Search for error messages containing the term opentelemetry. These error message might reveal problems similar to the following example, which exposes a misconfigured Collector endpoint:

2021-03-24 13:01:28.597  INFO 1 --- [           main]
o.hibernate.annotations.common.Version   : HCANN000001: Hibernate
Commons Annotations {5.1.0.Final}

[ 2021-03-24 13:01:29:072 +0000]
[BatchSpanProcessor_WorkerThread-1] WARN
- Failed to export spans

io.jaegertracing.internal.exceptions.SenderException: Could not send 16 spans


... stacktrace cut for brevity ...


2021-03-24 13:01:29.831  INFO 1 --- [           main]
org.hibernate.dialect.Dialect            : HHH000400: Using
dialect: org.hibernate.dialect.MySQL57Dialect

If the preceding steps don’t help you resolve your instrumentation issues, contact Splunk Observability Cloud support.