Folzo
08cc05987c
- diagnose-gpu.sh: Complete diagnostic script for GPU usage - GPU-HIGH-USAGE-FIX.md: Solution for OpenWebUI green icon causing 100% GPU - Issue #21062 confirmed: animate-ping CSS animation is the culprit - Multiple solutions provided (CSS, browser extension, zoom)
74 lines
2.2 KiB
Bash
Executable File
74 lines
2.2 KiB
Bash
Executable File
#!/bin/bash
|
|
# Script de diagnostic GPU quand utilisation à 95-100%
|
|
# À exécuter sur folzo-Corp quand tu vois le GPU tourner à fond
|
|
|
|
echo "=========================================="
|
|
echo "DIAGNOSTIC GPU - $(date)"
|
|
echo "=========================================="
|
|
echo ""
|
|
|
|
# 1. Utilisation GPU actuelle
|
|
echo "1. UTILISATION GPU :"
|
|
echo "--------------------"
|
|
rocm-smi
|
|
echo ""
|
|
|
|
# 2. Processus utilisant le GPU
|
|
echo "2. PROCESSUS GPU :"
|
|
echo "------------------"
|
|
rocm-smi --showpids
|
|
echo ""
|
|
|
|
# 3. Détails des processus Python (souvent les coupables)
|
|
echo "3. PROCESSUS PYTHON :"
|
|
echo "---------------------"
|
|
ps aux | grep -E "(python|ollama|open-webui)" | grep -v grep
|
|
echo ""
|
|
|
|
# 4. Pour chaque PID utilisant le GPU, voir la ligne de commande complète
|
|
echo "4. DÉTAILS DES PROCESSUS GPU :"
|
|
echo "------------------------------"
|
|
for pid in $(rocm-smi --showpids 2>/dev/null | grep -E "^[0-9]" | awk '{print $1}'); do
|
|
echo "PID: $pid"
|
|
echo "Commande: $(cat /proc/$pid/cmdline 2>/dev/null | tr '\0' ' ')"
|
|
echo "Utilisateur: $(ps -o user= -p $pid 2>/dev/null)"
|
|
echo "CPU: $(ps -o %cpu= -p $pid 2>/dev/null)%"
|
|
echo "Mémoire: $(ps -o %mem= -p $pid 2>/dev/null)%"
|
|
echo "---"
|
|
done
|
|
echo ""
|
|
|
|
# 5. Threads actifs des processus importants
|
|
echo "5. THREADS ACTIFS :"
|
|
echo "-------------------"
|
|
ps aux | grep -E "(python|ollama)" | grep -v grep | awk '{print $2}' | while read pid; do
|
|
echo "PID $pid: $(cat /proc/$pid/cmdline 2>/dev/null | tr '\0' ' ' | cut -c1-80)"
|
|
ls /proc/$pid/task/ 2>/dev/null | wc -l | xargs echo " Threads:"
|
|
done
|
|
echo ""
|
|
|
|
# 6. Connexions réseau actives
|
|
echo "6. CONNEXIONS RÉSEAU :"
|
|
echo "---------------------"
|
|
sudo netstat -tulpn 2>/dev/null | grep -E "(ollama|python|8188|11434)" || echo "Aucune connexion trouvée"
|
|
echo ""
|
|
|
|
# 7. Utilisation CPU/Mémoire globale
|
|
echo "7. UTILISATION SYSTÈME :"
|
|
echo "------------------------"
|
|
echo "CPU:"
|
|
top -bn1 | grep "Cpu(s)" | head -1
|
|
echo ""
|
|
echo "Mémoire:"
|
|
free -h | grep -E "(Mem|Swap)"
|
|
echo ""
|
|
|
|
# 8. Sauvegarde dans un fichier
|
|
LOG_FILE="/tmp/gpu-diagnostic-$(date +%Y%m%d-%H%M%S).log"
|
|
echo "Sauvegarde dans: $LOG_FILE"
|
|
$0 > "$LOG_FILE" 2>&1
|
|
echo ""
|
|
echo "=========================================="
|
|
echo "Diagnostic terminé !"
|
|
echo "=========================================="
|