{"id":6671,"date":"2026-06-08T13:14:39","date_gmt":"2026-06-08T11:14:39","guid":{"rendered":"https:\/\/roccadomenico.it\/wordpress\/?p=6671"},"modified":"2026-06-08T13:18:14","modified_gmt":"2026-06-08T11:18:14","slug":"ollama-su-docker","status":"publish","type":"post","link":"https:\/\/roccadomenico.it\/wordpress\/ollama-su-docker\/","title":{"rendered":"Ollama su docker"},"content":{"rendered":"\t\t
\n\t\t\t\t
\n\t\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t
\n\t\t\t\t\t\t\t\t\t

1. Prerequisiti<\/h2>

Verifica che i driver NVIDIA siano installati:<\/p>

nvidia-smi<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Dovresti vedere la tua GPU e la versione del driver.<\/p>
2. Installa Docker<\/h2>
sudo<\/span> apt update
sudo<\/span> apt install -y<\/span> docker.io docker-compose-plugin
sudo<\/span> systemctl enable docker
sudo<\/span> systemctl start<\/span> docker<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Aggiungi il tuo utente al gruppo docker:<\/p>
sudo<\/span> usermod -aG<\/span> docker $USER<\/span><\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Poi esegui logout\/login.<\/p>
3. Installa NVIDIA Container Toolkit<\/h2>
Aggiungi il repository:<\/p>
curl<\/span> -fsSL<\/span> https:\/\/nvidia.github.io\/libnvidia-container\/gpgkey | \\
sudo<\/span> gpg --dearmor<\/span> -o<\/span> \/usr\/share\/keyrings\/nvidia-container-toolkit-keyring.gpg

curl<\/span> -s<\/span> -L<\/span> https:\/\/nvidia.github.io\/libnvidia-container\/stable\/deb\/nvidia-container-toolkit.list | \\
sed<\/span> 's#deb https:\/\/#deb [signed-by=\/usr\/share\/keyrings\/nvidia-container-toolkit-keyring.gpg] https:\/\/#g'<\/span> | \\
sudo<\/span> tee<\/span> \/etc\/apt\/sources.list.d\/nvidia-container-toolkit.list<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Installa:<\/p>
sudo<\/span> apt update
sudo<\/span> apt install -y<\/span> nvidia-container-toolkit<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Configura Docker:<\/p>
sudo<\/span> nvidia-ctk runtime configure --runtime<\/span>=<\/span>docker
sudo<\/span> systemctl restart<\/span> docker<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Test:<\/p>
docker run --rm<\/span> --gpus<\/span> all ubuntu nvidia-smi<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Se visualizzi la GPU, sei pronto.<\/p>
4. Crea la cartella di Ollama<\/h2>
mkdir<\/span> -p<\/span> ~\/ollama
cd<\/span> ~\/ollama<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
5. Crea il file docker-compose.yml<\/h2>
services:
  ollama:
    image: ollama\/ollama:latest
    container_name: ollama

    ports:
      - \"11434:11434\"<\/span>

    volumes:
      - ollama_data:\/root\/.ollama

    restart: unless-stopped

    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

volumes:
  ollama_data:<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
6. Avvia Ollama<\/h2>
docker compose up -d<\/span><\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Verifica:<\/p>
docker ps<\/span><\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
7. Scarica un modello<\/h2>
Ad esempio Qwen 2.5 7B:<\/p>
docker exec -it<\/span> ollama ollama pull qwen2.5:7b<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Oppure:<\/p>
docker exec -it<\/span> ollama ollama pull qwen3:8b<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
8. Verifica che usi la GPU<\/h2>
Lancia:<\/p>
docker exec -it<\/span> ollama ollama run qwen2.5:7b<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
In un altro terminale:<\/p>
watch -n<\/span> 1<\/span> nvidia-smi<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Dovresti vedere memoria GPU occupata e utilizzo della GPU<\/p>
comandi Ollama sotto docker<\/p>
Se Ollama \u00e8 in esecuzione dentro un container Docker chiamato ollama<\/code>, puoi vedere tutti i modelli installati con:<\/p>
docker exec -it<\/span> ollama ollama list

<\/code><\/pre>
Rimuovere un modello<\/strong><\/p>
docker exec -it<\/span> ollama ollama rm<\/span> qwen2.5:7b<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Entrare nella shell del container<\/strong><\/div><\/div><\/div><\/div><\/div><\/div>
docker exec -it<\/span> ollama bash<\/span><\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
e poi eseguire direttamente:<\/div><\/div><\/div><\/div><\/div><\/div>
ollama list<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
Se vuoi verificare anche quanto spazio occupano i modelli nel volume Docker:<\/div><\/div><\/div><\/div><\/div><\/div>
docker system df -v<\/span><\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
oppure:<\/div><\/div><\/div><\/div><\/div><\/div>
docker exec -it<\/span> ollama du -sh<\/span> \/root\/.ollama\/models<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>
<\/code><\/pre><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div><\/div>\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"
1. Prerequisiti Verifica che i driver NVIDIA siano installati: nvidia-smi Dovresti vedere la tua GPU e la versione del driver. 2. Installa Docker sudo apt updatesudo apt install -y docker.io docker-compose-pluginsudo systemctl enable dockersudo systemctl start docker Aggiungi il tuo utente al gruppo docker: sudo usermod -aG docker $USER Poi esegui logout\/login. 3. Installa NVIDIA […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_post_was_ever_published":false},"categories":[107,30],"tags":[],"class_list":["post-6671","post","type-post","status-publish","format-standard","hentry","category-a-i","category-informatica"],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/posts\/6671","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/comments?post=6671"}],"version-history":[{"count":4,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/posts\/6671\/revisions"}],"predecessor-version":[{"id":6675,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/posts\/6671\/revisions\/6675"}],"wp:attachment":[{"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/media?parent=6671"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/categories?post=6671"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/roccadomenico.it\/wordpress\/wp-json\/wp\/v2\/tags?post=6671"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}