[Bug]: Solution to M4 mac arch64 elastic container crashing and $PY rag/svr/task_executor.py $1 issue with main server.

[B1tMaster]

Is there an existing issue for the same bug?

• [x] I have checked the existing issues.

RAGFlow workspace code commit ID

4e2afcd

RAGFlow image version

nightly

Other environment information

M4, macmini pro , 15.3.1 (24D70)

Actual behavior

this is actual solution to solve the problem.

I added below 3 ENV variables to my .env file under docker directory:

_JAVA_OPTIONS=-XX:UseSVE=0 OPENBLAS_CORETYPE=generic NUMBA_DISABLE_JIT=1

This solution solves 2 problems:

1. Elastic search container constantly re-boot due to Java crash.. related to MacBook m4 macmini ARch64 incompatibility .
2. main ragflow server , not being able to start properly due to executor failing all the time again due to MacBook m4 Arch64 incompatibility .
   basically if you get this: illegal $PY rag/svr/task_executor.py $1

Expected behavior

No response

Steps to reproduce

this is a fix.. just add those lines to .env file if you are on macOS m4 

_JAVA_OPTIONS=-XX:UseSVE=0
OPENBLAS_CORETYPE=generic
NUMBA_DISABLE_JIT=1

Additional information

No response

[KevinHuSh]

Also, you could build an image from source.

git clone https://github.com/infiniflow/ragflow.git
cd ragflow/
pip3 install huggingface_hub nltk
python3 download_deps.py
docker build -f Dockerfile.deps -t infiniflow/ragflow_deps .
docker build -f Dockerfile -t infiniflow/ragflow:nightly .
docker build --build-arg LIGHTEN=1 -f Dockerfile -t infiniflow/ragflow:nightly-slim .

[jt-z]

I used the two solutions above, set the environment variables in the env, and pulled the nightly code to rebuild the image, and finally it could run on the M4 chip.

[rollinsio]

I ended up using this, but also had the issue that elastic search was set to use 8GB memory. However, I had only allotted 6GB for all of Docker. Once I bumped ES down to 4GB (which is still a lot) things started running smoothly