WEBDIENSTE/ai-suite
1
0
Fork 0
mirror of https://github.com/trevorsandy/ai-suite.git synced 2026-09-30 13:35:13 +02:00
Code Issues Packages Projects Releases Wiki Activity

Refactor llama process to use localhost domain when running on host

Browse Source
This commit is contained in:
Trevor SANDY
2026-05-15 00:20:45 +02:00
parent e4145b328d
commit 2086c09ba3
2 changed files with 24 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.env.example
+6 -8
View File
@@ -374,13 +374,12 @@ MIN_DOCUMENT_SIZE_FOR_CHUNKING=100000 # Only chunk very large documents
OLLAMA_PORT=11434
# Backend connect when running Ollama in the Host:
# host.docker.internal:11434
# When running Ollama in the Host:
#OLLAMA_HOST=localhost:11434
# Docker backend connect when running Ollama in the Host:
#OLLAMA_HOST=host.docker.internal:${OLLAMA_PORT}
# When accessing Ollama from the Host:
OLLAMA_HOST=localhost:${OLLAMA_PORT}
# When running Ollama in Docker:
#OLLAMA_HOST=ollama:11434
OLLAMA_HOST=localhost:11434
#OLLAMA_HOST=ollama:${OLLAMA_PORT}
# Tuning
OLLAMA_CONTEXT_LENGTH=4096
@@ -402,10 +401,9 @@ OLLAMA_SERVER_ARGS=serve
LLAMA_ARG_PORT=8040
# When running LLaMA.cpp in the host:
# Docker backend connect when running LLaMA.cpp in the Host:
#LLAMA_ARG_HOST=host.docker.internal
# When running LLaMA.cpp in Docker:
#LLAMA_ARG_HOST=0.0.0.0
LLAMA_ARG_HOST=0.0.0.0
# Backend connect
suite_services.py
+18 -5
View File
@@ -1,7 +1,7 @@
#!/usr/bin/env python3
"""
Trevor SANDY
Last Update April 04, 2026
Last Update April 09, 2026
Copyright (c) 2025-Present by Trevor SANDY
AI-Suite uses this script for the installation command that handles the AI-Suite
@@ -1303,8 +1303,9 @@ def _docker_test_container():
return True
return False
def launch_llama_process(args, llama_log):
def launch_llama_process(args, env=None, llama_log=None):
"""Launch Ollama/LLaMA.cpp server on the host"""
llama_log = "llama_start.log" if not llama_log else llama_log
log_file = "".join(['>', llama_log, ' 2>&1'])
if system == "Windows":
win = "".join(['/c,"', llama_exe])
@@ -1315,7 +1316,13 @@ def launch_llama_process(args, llama_log):
raw_msg = " ".join([log_run_cmd, " ".join(cmd)])
log.info(raw_msg, extra=LSHF.style(header=log_run_cmd, msg=" ".join(cmd)))
try:
completed = subprocess.run(cmd, capture_output=True, text=True, check=True)
completed = subprocess.run(
cmd,
capture_output=True,
env=env,
text=True,
check=True
)
if completed.returncode != 0:
log.error(f"Command: {llama} process: {completed.stderr}")
except Exception as e:
@@ -1511,9 +1518,15 @@ def check_llama_process(operation=None, env_vars={}):
llama_server_args = env_vars.get('OLLAMA_SERVER_ARGS')
if llama_server_args:
llama_args.extend([llama_server_args])
llama_host = "localhost"
llama_port = env_vars.get('OLLAMA_PORT')
llama_host_var = llama_host if llama_cpp else f"{llama_host}:{llama_port}"
llama_host_env = "LLAMA_ARG_HOST" if llama_cpp else "OLLAMA_HOST"
log.info(f"Set '{llama_host_env}' to '{llama_host_var}' in subprocess env...")
env = os.environ.copy()
env[llama_host_env] = llama_host_var
args = " ".join(llama_args)
launch_llama_process(args, llama_log_file)
launch_llama_process(args, env, llama_log_file)
else:
log.critical(f"The {llama_app} file was not found at {llama_exe}.")
log.critical(f"If {llama} is installed in a non-standard location, set the LLAMA_PATH")