How do you configure server logs? Looks like it won't log to a file no matter what the settings

[elabz]

Hello all,
This issue baffles me for several hours, looks like I need help here. I am trying to configure llama.cpp server to run in a Docker container, which in and of itself is not very difficult to do. However, there does not seem to be a way to make it output the logs anywhere but stdout

I am starting the server as
llama-server --model /models/L3-8B-Stheno-v3.2-Q5_K_M-imat.gguf --logdir /logs --log-file llama --ctx-size 4096 --n-gpu-layers 99 --host 0.0.0.0 --log-enable --log-new --log-append

so you can see I have thrown every log related setting on ON, still nothing. the /logs directory is mapped in docker-compose.yml to a local directory as such:

   ...
    volumes:
      - ./models:/models
      - ./logs:/logs

I know this isn't a Docker permission issue because I am running a script in the same container that downloads models, and is able to save them to /models, which is mapped in the exact same way right there. So the issue seems to be the configuration?
Am I not calling the server correctly?
Thanks for your ideas!

P.S. Is there a way to capture prompts and outputs in a log? Would love to have that feature for prompt tuning!

[red-co]

Hello, is this problem solved?
I wasted several hours, and I found that llama-server will not write any logs no matter what, it will always be "warming up the model with an empty run", but llama-cli can log normally.
Is it my problem? Or does llama-server not support logs?
If it does, is there a simplest example?

[zijinyuan]

me too, the same problem, can't dump any log info, try any -log parameters.