dongho/Munich-news
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update

Browse Source
This commit is contained in:
Dongho Kim
2025-11-12 15:34:19 +01:00
parent d09d96ca4b
commit 3a6a36b636
1 changed files with 26 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
backend/routes/ollama_routes.py
View File

@@ -221,19 +221,41 @@ def get_gpu_status():
import subprocess
# Check if nvidia-smi works in ollama container
result = subprocess.run(
['docker', 'exec', 'munich-news-ollama', 'nvidia-smi', '--query-gpu=name', '--format=csv,noheader'],
['docker', 'exec', 'munich-news-ollama', 'nvidia-smi', '--query-gpu=name,memory.used,utilization.gpu', '--format=csv,noheader'],
capture_output=True,
text=True,
timeout=5
)
if result.returncode == 0 and result.stdout.strip():
gpu_data = result.stdout.strip().split(',')
gpu_info['gpu_available'] = True
gpu_info['gpu_details'] = {
'gpu_name': result.stdout.strip(),
'gpu_name': gpu_data[0].strip() if len(gpu_data) > 0 else 'Unknown',
'memory_used': gpu_data[1].strip() if len(gpu_data) > 1 else 'N/A',
'utilization': gpu_data[2].strip() if len(gpu_data) > 2 else 'N/A',
'note': 'GPU available but no model currently loaded'
}
except:
pass
# Check Ollama logs for GPU usage evidence
log_result = subprocess.run(
['docker', 'logs', '--tail', '50', 'munich-news-ollama'],
capture_output=True,
text=True,
timeout=5
)
if log_result.returncode == 0:
logs = log_result.stdout + log_result.stderr
# Look for GPU offloading messages
if 'offloaded' in logs.lower() and 'gpu' in logs.lower():
gpu_info['gpu_in_use'] = True
gpu_info['gpu_details']['note'] = 'GPU acceleration active (detected in logs)'
# Extract layer info if available
import re
match = re.search(r'offloaded (\d+)/(\d+) layers', logs, re.IGNORECASE)
if match:
gpu_info['gpu_details']['layers_offloaded'] = f"{match.group(1)}/{match.group(2)}"
except Exception as e:
gpu_info['debug_error'] = str(e)
# Try to get system info
try: