New Relic agent is installed but never attached to the JVM (v5.0.4 and later)
Los mantenedores suelen responder en 1 día
Evaluación
- Dificultad
- 3/5
- Tiempo estimado
- 1-2 días
- Aptitud para principiantes
- 35/100
- Tipo de issue
- Error
- Claridad
- Bien especificado
- Estado de actividad
- Estancado
- Área
- build-system, cloud
Línea de trabajo
Start with the New Relic framework code in new_relic.go and the 9.x entries in manifest.yml, which show the bare agent jar with a .zip name. The fix that matters is upstream in buildpacks-ci, where the watcher downloads the wrong artifact, and an open pull request (#714) already exists for this. Done means an integration test that checks newrelic.jar and the license key in the running JVM's arguments, and passes for New Relic 9.x.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
With a newrelic service binding, staging succeeds and the log says the agent is installed, but the running JVM has no New Relic -javaagent and no license key. The only hint is a warning during finalize:
WARNING Failed to finalize framework New Relic Agent: failed to locate newrelic.jar: newrelic.jar not found in /tmp/contents.../deps/0/new_relic_agent
Datadog and SkyWalking bound to the same app are present in the JVM arguments, New Relic is not.
How it was found: while writing an integration test that binds several agents at once (New Relic, Datadog, SkyWalking) and checks the running JVM's arguments instead of only the staging log. This is part of an upcoming improvement of the integration tests, so that every framework is verified in the running JVM. So far 17 of the 23 frameworks that attach an agent (-javaagent or -agentpath) have been checked this way: 16 pass and New Relic is the only one that fails. The other 6 are not checked: AppDynamics, Checkmarx, Introscope and Riverbed are skipped because they need vendor credentials, and Seeker and Java Memory Assistant have no test.
Cause: new_relic.go expects newrelic/newrelic.jar inside the unpacked dependency, as in the vendor distribution newrelic-java-<version>.zip (used up to 8.15.0). The 9.x dependencies in manifest.yml (9.2.0 to 9.5.0, first shipped in v5.0.4) are the bare agent jar with a .zip name. libbuildpack unpacks .zip dependencies, so the jar is exploded and no newrelic.jar remains.
The pipeline produced this: the newrelic watcher in buildpacks-ci downloads newrelic-agent-<version>.jar (cloudfoundry/buildpacks-ci#633), while the binary-builder recipe names the artifact newrelic-java-<version>.zip (cloudfoundry/binary-builder#111). The recipe never opens the file, so the .zip extension comes from its expected file name, not from the content.
Fix: make the watcher download newrelic-java-<version>.zip. The distribution exists for 9.2.0, 9.4.0 and 9.5.0 and contains newrelic/newrelic.jar (the same 41,319,404 bytes as the published bare jar for 9.5.0). I plan to open that PR in buildpacks-ci, plus an integration test here that asserts newrelic.jar and the license key in the running JVM's arguments. The corrected artifact is published when the pipeline next builds a New Relic version; I have not checked whether 9.5.0 can be rebuilt on demand.
Why it went unnoticed: the integration test only asserts New Relic Agent in the staging log, which is printed before the failing step. A failed Finalize is only a warning, while Supply failures are fatal. Making Finalize failures fatal is a separate behaviour change worth discussing.
v4: the Ruby buildpack used a bare jar (download_jar) and added -javaagent directly, so this worked.
- Lenguaje dominante
- Go
- Estrellas
- 455
- Forks
- 2.5k
- Merge medio
- 22 h 9 min
- PR fusionados (30 d)
- 69
Preparar el entorno
- Sin Dockerfile ni archivo de Docker Compose
- Sin plantilla de pull request
- Leer la guía de contribución
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de cloudfoundry/java-buildpack
-
Dificultad 5/5 Más de una semana Aptitud para principiantes 12/100
cloudfoundry/java-buildpack#1549 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 5/5 Más de una semana Aptitud para principiantes 25/100
cloudfoundry/java-buildpack#1521 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
Dificultad 5/5 Más de una semana Aptitud para principiantes 10/100
cloudfoundry/java-buildpack#1469 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
Support configurable initial/minimum heap size relative to (/as a percentage of) calculated XmxAbierto
Dificultad 3/5 1-2 días Aptitud para principiantes 55/100
cloudfoundry/java-buildpack#1377 · 5 comentarios ·
Los mantenedores suelen responder en 1 día
-
Dificultad 3/5 1-2 días Aptitud para principiantes 52/100
cloudfoundry/java-buildpack#1364 ·
Los mantenedores suelen responder en 1 día
Todos los issues de cloudfoundry/java-buildpack
Issues similares
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 84/100
github/github-mcp-server#3865 ·
Los mantenedores suelen responder en 2 días
-
kind/features
Dificultad 2/5 1-3 horas Aptitud para principiantes 75/100
TencentBlueKing/bk-nodemgr#4017 ·
Los mantenedores suelen responder en 1 día
-
ssa: wait errors drop the kstatus message, so health check failures don't say whyPosiblemente ocupada @schdief la tomó hoy. Abierto
Dificultad 2/5 1-3 horas Aptitud para principiantes 76/100
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 76/100
equinor/radix-operator#2066 ·
Los mantenedores suelen responder en 2 días
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 67/100
microsoft/terraform-provider-msgraph#188 · 1 reacción ·
Los mantenedores suelen responder en 1 día