Debugging with Skaffold

This page describes skaffold debug, a zero-configuration solution for setting up containers for debugging on a Kubernetes cluster.

Debugging with Skaffold

skaffold debug acts like skaffold dev, but it configures containers in pods for debugging as required for each container’s runtime technology. The associated debugging ports are exposed and labelled so that they can be port-forwarded to the local machine. Helper metadata is also added to allow IDEs to detect the debugging configuration parameters.

How it works

skaffold debug examines the built artifacts to determine the underlying runtime technology. Any Kubernetes manifest that references these artifacts are transformed to enable the runtime technology’s debugging functions.

skaffold debug uses a set of heuristics to identify the runtime technology.
The Kubernetes manifests are transformed on-the-fly such that the on-disk representations are untouched.

Each Pod will have an annotation with a JSON object describing the debug configurations for the pod’s containers (linebreaks for readability):{

For example the following annotation indicates that the container named web is a Go application that is being debugged by a headless Delve session on port 56268:{"web":{"dlv":56268,"runtime":"go"}}

Some language runtimes require additional support files to enable debugging. For these languages, a special set of runtime-specific images are configured as init-containers to populate a shared-volume that is mounted into each of the appropriate containers. These images are hosted at

Supported Language Runtimes

Debugging is currently supported for Go, Java (and JVM languages), NodeJS, and Python.


Go-based applications are configured to run under Delve in its headless-server mode.

  • Go application should self-identify by setting one of the standard Go runtime environment variables such as GODEBUG, GOGC, GOMAXPROCS, or GOTRACEBACK. GOTRACEBACK=all is a generally useful configuration.
  • Go applications should be built without optimizations, so your build should be capable of building with -gcflags='all=-N -l'. Skaffold Profiles are a useful option.

Note for users of VS Code’s debug adapter for Go: Delve seems to treat the source location for headless launches as being relative to /go. The following remote launch configuration was useful:

  "name": "Skaffold Debug",
  "type": "go",
  "request": "launch",
  "mode": "remote",
  "host": "localhost",
  "port": 56268,
  "remotePath": "/go/",
  "program": "/Users/login/go/src/",

Java and other JVM languages

Java/JVM applications are configured to expose the JDWP agent using the JAVA_TOOL_OPTIONS environment variable.
Note that the use of JAVA_TOOL_OPTIONS causes extra debugging output from the JVM on launch.


NodeJS applications are configured to use the Chrome DevTools inspector.
NodeJS applications must be launched using node or nodemon, or npm. Note that npm scripts should not then invoke nodemon as the DevTools inspector configuration will be picked up by nodemon rather than the actual application.

Note that the client must first obtain the inspector UUID.


Python applications are configured to use ptvsd, a wrapper around pydevd that uses the debug adapter protocol (DAP).

The DAP is supported by Visual Studio Code, Eclipse LSP4e, and other editors and IDEs. DAP is not yet supported by JetBrains IDEs like PyCharm.


skaffold debug has some limitations:

  • Only the kubectl and kustomize deployers are supported at the moment: support for the Helm deployer is not yet available.
  • File sync is disabled for all artifacts.