Code Review

Review: Refactor dmr and extract server package

Summary

This PR refactors the dmr tool to use cobra and migrates the core daemon logic into a reusable pkg/server package. While the modularization is a positive step, there are critical issues with error handling and a missing variable declaration that will prevent successful compilation and operation.

Critical

• main.go:20: The Version variable is used but not declared in the package, which will cause a linker error.
• pkg/server/server.go:186: Calling os.Exit(1) inside a library package prevents graceful cleanup; errors should be returned to the caller.
• pkg/server/server.go:334: The Run function logs server errors but returns nil, masking fatal failures from the supervisor.

What looks good

• The extraction of server logic into a dedicated package improves modularity and testability.

Hey - I've left some high level feedback:

• The new Run function in pkg/server is quite large and mixes configuration, logging setup, backend wiring, and serving logic; consider breaking it into smaller helpers (e.g., logging setup, service construction, HTTP/TLS listener setup) to improve readability and make targeted behavior easier to test.
• OnBackendError currently calls ExitFunc directly, which makes the server package less reusable as a library; you might instead propagate backend init failures via a returned error (or a channel) so callers can decide how to handle fatal startup conditions.

Please address the comments from this code review:

## Overall Comments
- The new Run function in pkg/server is quite large and mixes configuration, logging setup, backend wiring, and serving logic; consider breaking it into smaller helpers (e.g., logging setup, service construction, HTTP/TLS listener setup) to improve readability and make targeted behavior easier to test.
- OnBackendError currently calls ExitFunc directly, which makes the server package less reusable as a library; you might instead propagate backend init failures via a returned error (or a channel) so callers can decide how to handle fatal startup conditions.