Isolating containers from internet access to enhance security.

9 files changed, 53 insertions, 22 deletions

diff --git a/.gitignore b/.gitignore
index 5f12fcd..61316fd 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 .env
 tools/aichat
+src/nginx/htpasswd
diff --git a/README.md b/README.md
index cc6a3e6..a442987 100644
--- a/README.md
+++ b/README.md
@@ -46,7 +46,7 @@ Add an API key to secure server access by adding a `.env` file like this:
 LLM_API_KEY=1234567890
 ```
 
-Create a user authentication for aichat web UI:
+Create a user authentication for aichat Web UI:
 
 ```
 htpasswd -c src/nginx/htpasswd user
@@ -60,7 +60,7 @@ docker compose up --build -d
 
 Then wait for the models to finish downloading using the following command to display the status:
 ```
-docker-compose logs -f llm_provision
+docker-compose logs -f ollama_provision
 ```
 
 ## How to use
@@ -120,3 +120,9 @@ Example:
 export TTS_API_HOST="https://your-remote-domain"
 ./tools/speech.sh ... 
 ```
+
+## Web UI
+
+A web application to interact with supported LLMs directly from your browser is available at [http://127.0.0.1:8000/playground](http://127.0.0.1:8000/playground).
+
+A web platform to compare different LLMs side-by-side is available at [http://127.0.0.1:8000/arena](http://127.0.0.1:8000/arena).
diff --git a/docker-compose.yml b/docker-compose.yml
index 65638a9..f750995 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -17,6 +17,8 @@ services:
       retries: 5
       start_period: 20s
       timeout: 10s
+    networks:
+      - nointernet
 
   openedai-speech:
     build:
@@ -40,12 +42,16 @@ services:
       retries: 5
       start_period: 10s
       timeout: 10s
+    networks:
+      - nointernet
 
-  llm_provision:
+  ollama_provision:
     build:
-      dockerfile: src/llm_provision/Dockerfile
+      dockerfile: src/ollama_provision/Dockerfile
     environment:
       - MODELS=qwen2.5:latest,qwen2.5-coder:32b,nomic-embed-text:latest,gemma2:latest,mistral:latest,deepseek-r1:7b
+    volumes:
+      - ollama:/root/.ollama
     restart: no
     depends_on:
       ollama:
@@ -53,6 +59,8 @@ services:
         restart: true
     links:
       - ollama
+    networks:
+      - internet
 
   aichat:
     build:
@@ -69,11 +77,14 @@ services:
       interval: 30s
       timeout: 15s
       retries: 3
+    networks:
+      - nointernet
 
   nginx:
     image: nginx
     volumes:
       - ./src/nginx/nginx.conf:/etc/nginx/templates/nginx.conf.template
+      - ./src/nginx/htpasswd:/etc/nginx/.htpasswd
     environment:
       - NGINX_ENVSUBST_OUTPUT_DIR=/etc/nginx
       - API_KEY=${LLM_API_KEY}
@@ -90,9 +101,17 @@ services:
       - "8000:8000"
       - "8001:8001"
     restart: unless-stopped
+    networks:
+      - internet
+      - nointernet
 
 volumes:
   ollama:
   voices:
   speech-config:
-  hf-hub-cache:
+
+networks:
+  internet:
+    internal: false
+  nointernet:
+    internal: true
diff --git a/src/aichat/Dockerfile b/src/aichat/Dockerfile
index 406dde2..a4d33bd 100644
--- a/src/aichat/Dockerfile
+++ b/src/aichat/Dockerfile
@@ -7,8 +7,15 @@ RUN update-ca-certificates
 RUN cargo install --target x86_64-unknown-linux-musl aichat
 
 ADD src/aichat/entrypoint.sh /entrypoint.sh
-ADD src/aichat/config.yaml /aichat_config_tpl.yaml
-
 RUN chmod 755 entrypoint.sh
 
+RUN useradd -ms /bin/bash aichat
+USER aichat
+WORKDIR /home/aichat
+
+RUN mkdir -p /home/aichat/.config/aichat
+
+ADD src/aichat/config.yaml /home/aichat/.config/aichat/config.yaml
+ADD src/aichat/roles /home/aichat/.config/aichat/roles
+
 ENTRYPOINT ["/entrypoint.sh"]
diff --git a/src/aichat/entrypoint.sh b/src/aichat/entrypoint.sh
index ec4f040..77d9285 100644
--- a/src/aichat/entrypoint.sh
+++ b/src/aichat/entrypoint.sh
@@ -1,4 +1,6 @@
 #!/bin/sh
-mkdir -p ~/.config/aichat
-cat /aichat_config_tpl.yaml | sed "s/__LLM_API_KEY__/${LLM_API_KEY}/" | sed "s/localhost/ollama/" >~/.config/aichat/config.yaml
+
+cat ~/.config/aichat/config.yaml | grep -v 'api_key' | sed "s/localhost/ollama/" | tee ~/.config/aichat/config.yaml.tmp
+mv ~/.config/aichat/config.yaml.tmp ~/.config/aichat/config.yaml
+
 aichat --serve 0.0.0.0
diff --git a/src/nginx/nginx.conf b/src/nginx/nginx.conf
index fa4cb13..f07765a 100644
--- a/src/nginx/nginx.conf
+++ b/src/nginx/nginx.conf
@@ -39,14 +39,10 @@ http {
     }
     server {
         listen 8001;
-        set $deny 1;
-        if ($http_authorization = "Bearer $API_KEY") {
-            set $deny 0;
-        }
-        if ($deny) {
-            return 403;
-        }
         location / {
+            auth_basic "Private Area";
+            auth_basic_user_file /etc/nginx/.htpasswd;
+
             proxy_pass http://aichat:8000;
             proxy_set_header Host $host;
             proxy_set_header X-Real-IP $remote_addr;
diff --git a/src/llm_provision/Dockerfile b/src/ollama_provision/Dockerfile
index 77701fe..4aa439b 100644
--- a/src/llm_provision/Dockerfile
+++ b/src/ollama_provision/Dockerfile
@@ -1,11 +1,11 @@
-FROM debian:bookworm-slim
+FROM ollama/ollama:latest
 
 ENV DEBIAN_FRONTEND=noninteractive
 RUN apt-get update
 RUN apt-get --yes -o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confnew" install bash curl jq
 
-ADD ./src/llm_provision/init_models.sh /init_models.sh
-ADD ./src/llm_provision/entrypoint.sh /entrypoint.sh
+ADD ./src/ollama_provision/init_models.sh /init_models.sh
+ADD ./src/ollama_provision/entrypoint.sh /entrypoint.sh
 RUN chmod 755 /entrypoint.sh
 
 ENTRYPOINT ["/entrypoint.sh"]
diff --git a/src/llm_provision/entrypoint.sh b/src/ollama_provision/entrypoint.sh
index d0b6e85..1952755 100644
--- a/src/llm_provision/entrypoint.sh
+++ b/src/ollama_provision/entrypoint.sh
@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
 
+ollama start &
 echo "pull models into ollama volumes"
 bash /init_models.sh
diff --git a/src/llm_provision/init_models.sh b/src/ollama_provision/init_models.sh
index 960eb98..1eae979 100755
--- a/src/llm_provision/init_models.sh
+++ b/src/ollama_provision/init_models.sh
@@ -1,17 +1,16 @@
 #!/usr/bin/env bash
 
-OLLAMA_HOST="http://ollama:11434"
 
 IFS=',' read -r -a models_arr <<< "${MODELS}"
 
 ## now loop through the above array
 for m in "${models_arr[@]}"
 do
-  curl -s "${OLLAMA_HOST}/api/tags" | jq '.models[].name' | grep ${m} > /dev/null
+  ollama list | tail -n +2 | cut -d' ' -f1 | grep ${m} > /dev/null
   if [[ $? -ne 0 ]]
   then
     echo "download {m}"
-    curl -s "${OLLAMA_HOST}/api/pull" -d "{\"model\": \"${m}\"}"
+    ollama pull "${m}"
   else
     echo "${m} already installed"
   fi