1. Build
   cd Gemini/src
   make CUDA_PATH=/usr/local/cuda-12.1/ PREFIX=/home/zhen1.zhang/2024_NOCUDA/Gemini/build DEBUG=1

2. launch-backend
   cd Gemini
   python tools/launch-backend.py --ip 127.0.0.1
   --port 40051
   --config ./resource-config.txt
   --base_quota 300
   --min_quota 20
   --window 10000
   ./build/bin/gem-schd
   ./build/bin/gem-pmgr

[launcher] scheduler started on 127.0.0.1:40051
[launcher] pod manager of client1 started on 127.0.0.1:40052
[launcher] pod manager of client2 started on 127.0.0.1:40053
2024-10-19 17:10:05.976263 Gemini I/ There are 2 clients in the system...
2024-10-19 17:10:05.976304 Gemini I/ client1 request: 0.50, limit: 0.50, memory limit: 327374182400 bytes
2024-10-19 17:10:05.976316 Gemini I/ client2 request: 0.50, limit: 0.50, memory limit: 327374182400 bytes
2024-10-19 17:10:05.976412 Gemini I/ Monitor thread created.
2024-10-19 17:10:05.976421 Gemini I/ Waiting for incoming connection
2024-10-19 17:10:05.976438 Gemini E/ Failed to add watch to '.'.
2024-10-19 17:10:05.976558 Gemini I/ Received an incoming connection.
2024-10-19 17:10:05.977031 Gemini I/ Received an incoming connection.

I have the Error: Failed to add watch to '.'.

I am not sure what is the reason for this error log.