7a40ff629e
Some checks failed
Build and Deploy GooSeek / build-and-deploy (push) Failing after 8m25s
- Add tier-based provider routing in llm-svc - free tier → Ollama (local qwen3.5:9b) - pro/business → Timeweb Cloud AI - Add /api/v1/embed endpoint for embeddings via Ollama - Update Ollama client: qwen3.5:9b default, remove auth - Add GenerateEmbedding() function for qwen3-embedding:0.6b - Add Ollama K8s deployment with GPU support (RTX 4060 Ti) - Add monitoring stack (Prometheus, Grafana, Alertmanager) - Add Grafana dashboards for LLM and security metrics - Update deploy.sh with monitoring and Ollama deployment Made-with: Cursor
39 lines
1.1 KiB
Bash
Executable File
39 lines
1.1 KiB
Bash
Executable File
#!/bin/bash
|
||
# Скрипт для загрузки моделей в Ollama
|
||
# Запустить ОДИН РАЗ после первого деплоя
|
||
# Модели сохраняются в PVC и не нужно скачивать повторно
|
||
|
||
set -e
|
||
|
||
NAMESPACE="${NAMESPACE:-gooseek}"
|
||
MODELS="${@:-llama3.2:3b}"
|
||
|
||
echo "=== Ollama Model Loader ==="
|
||
echo "Namespace: $NAMESPACE"
|
||
echo "Models: $MODELS"
|
||
|
||
# Проверить что Ollama pod запущен
|
||
echo ""
|
||
echo "Checking Ollama pod status..."
|
||
kubectl -n $NAMESPACE wait --for=condition=ready pod -l app=ollama --timeout=120s
|
||
|
||
# Получить имя пода
|
||
POD=$(kubectl -n $NAMESPACE get pod -l app=ollama -o jsonpath='{.items[0].metadata.name}')
|
||
echo "Pod: $POD"
|
||
|
||
# Скачать модели
|
||
for MODEL in $MODELS; do
|
||
echo ""
|
||
echo "=== Pulling model: $MODEL ==="
|
||
kubectl -n $NAMESPACE exec -it $POD -c ollama -- ollama pull $MODEL
|
||
done
|
||
|
||
# Показать список моделей
|
||
echo ""
|
||
echo "=== Installed models ==="
|
||
kubectl -n $NAMESPACE exec -it $POD -c ollama -- ollama list
|
||
|
||
echo ""
|
||
echo "=== Done! ==="
|
||
echo "Models are stored in PVC and will persist across restarts."
|