Summary

Fixes #2257.

When default_execution_mode: distributed is set and a parent DAG without worker_selector uses dag.run to invoke a sub-DAG, the parent gets dispatched to a remote worker. The worker's local DAG store does not contain the parent's sub-DAG definitions, causing rCtx.DB.GetDAG() to either be called on a nil DB or to return a nil DAG — both leading to nil pointer dereference in newSubDAGExecutor and crashing the worker pod.

Changes

internal/runtime/executor/dag_runner.go:

• Add nil-check on rCtx.DB before calling GetDAG
• Add nil-check on the returned dag value
• Both new error paths wrap exec.ErrDAGNotFound and include a remediation hint pointing users to set worker_selector: local on the parent DAG

internal/runtime/executor/dag_runner_test.go:

• New test TestNewSubDAGExecutor_NilDB — asserts structured error when rCtx.DB == nil
• New test TestNewSubDAGExecutor_NilDAGReturn — asserts structured error when GetDAG returns nil dag

Behaviour Change

Before:
```
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation]
```

After:
```
cannot resolve sub-DAG "child": no local DAG store available
(hint: parent DAG was dispatched to a worker without local DAG cache
— consider setting worker_selector: local on the parent DAG):
DAG is not found
```

Test Plan

```
go test ./internal/runtime/executor/...
ok github.com/dagucloud/dagu/internal/runtime/executor 0.015s
```

Local verification on a multi-cluster K3s setup confirmed the worker pod no longer crashes when an orchestrator DAG is mis-dispatched; the descriptive error appears in worker logs instead.

Related

See #2257 for the full reproduction scenario and background. Example DAGs demonstrating the pattern are documented in that issue's comments; happy to send a follow-up PR adding them to examples/embedded/distributed/ if the maintainers prefer.

Summary by cubic

Fixes a nil pointer panic when a parent DAG on a remote worker calls a sub-DAG missing from the worker’s local store. Now returns a clear “DAG not found” error with guidance, preventing worker crashes.

• Bug Fixes
  • Add nil checks on rCtx.DB and the GetDAG result in NewSubDAGExecutor.
  • Return errors wrapping exec.ErrDAGNotFound with a hint to use worker_selector: local on the parent DAG in distributed mode.
  • Add tests for nil DB and nil DAG cases, with doc comments; worker now logs a structured error instead of panicking.

^{Written for commit fba216b. Summary will update on new commits.}

Summary by CodeRabbit

• Bug Fixes

  • Improved validation when executing sub-workflows to prevent errors from missing database configurations.
  • Enhanced error messages to provide clearer guidance on required configuration settings for sub-workflow execution.

• Tests

  • Added comprehensive test coverage for error handling in sub-workflow executor scenarios.