Build fails: Failed to copy server binary to output directory: No such file or directory (os error 2)

[yurivict]

Describe the bug

 -- Installing: /wrkdirs/usr/ports/devel/tabby/work/target/release/build/llama-cpp-server-8837603d1835d022/out/bin/llama-tokenize
  cargo:root=/wrkdirs/usr/ports/devel/tabby/work/target/release/build/llama-cpp-server-8837603d1835d022/out

  --- stderr
  CMake Warning at cmake/build-info.cmake:14 (message):
    Git not found.  Build info will not be accurate.
  Call Stack (most recent call first):
    CMakeLists.txt:77 (include)


  CMake Warning at ggml/src/CMakeLists.txt:274 (message):
    AMX requires gcc version > 11.0.  Turning off GGML_AMX.


  CMake Warning at common/CMakeLists.txt:30 (message):
    Git repository not found; to enable automatic generation of build info,
    make sure Git is installed and the project is a Git repository.


  CMake Warning:
    Manually-specified variables were not used by the project:

      CMAKE_ASM_COMPILER
      CMAKE_ASM_FLAGS


  thread 'main' panicked at crates/llama-cpp-server/build.rs:66:36:
  Failed to copy server binary to output directory: No such file or directory (os error 2)

Information about your version
0.20.0

Additional context
FreeBSD 14.1

[zwpaper]

Hi @yurivict, it looks like the llama-server is not successfully built, can you provide some more details, like

1. the command you used
2. the full logs output

so that we can dig deeper into it

[yurivict]

Here is a complete log.

Also: we already have llama-cpp package available, so there should be no need to bundle it in tabby.
Is it possible to use the external llama-cpp package?

[wsxiaoys]

Also: we already have llama-cpp package available, so there should be no need to bundle it in tabby.

Yes - you can turn off the llama-cpp binary building by disabling this feature:

tabby/crates/llama-cpp-server/Cargo.toml

Line 9 in 39c5b8d

binary = []