OmniMES installation

The complete OmniMES deployment procedure - from downloading Docker, through license activation, Redash and the Ollama AI server configuration, to changing passwords, updating and uninstalling.

1. Prerequisites

Before installation, make sure the following ports are free on the target computer:

Docker is also required: https://docs.docker.com/get-docker/

2. Step-by-step installation

1. Download and install the Docker version appropriate for your system.

2. From the installation package, run the script matching your operating system:

   • Windows: install_all_windows.bat

   • Linux: install_all_linux.sh

   These scripts install the full version of OmniMES together with the components: API, Stream, WWW, Redash and Ollama.

3. Linux - granting permissions (before the first run):

   chmod +x install_linux.sh uninstall.sh updated_api.sh updated_www.sh
   ./install_all_linux.sh

4. Windows - the "Windows protected your PC" message Click More info → Run anyway.

3. First login

3.1. Opening the application in a browser

After the script finishes, open a browser and type one of the following addresses into the address bar:

• localhost

• 127.0.0.1

• the local IP address of the computer where the software was installed

The OmniMES login screen will appear.

3.2. Filling in the login screen

The login form has three fields - fill them in as follows:

ℹ️ Common mistake: in the API URL field you enter the address of the backend service (port 8091), not the address from which you open the browser (port 80). These addresses usually differ by port and may point to different hosts (e.g. the frontend locally, the API on a server).

⚠️ Security: The factory password should be changed immediately after the first login.

4. License activation

After logging in, the system automatically takes you to the activation screen. Enter the license key provided with the package - the product is ready to use.

5. Redash configuration (reports)

Redash is the built-in OmniMES reporting engine, available locally at:

http://localhost:5000/

ℹ️ After installation the Redash database is restored automatically - together with a ready-made data source, queries and dashboards. There is no need to manually add a JSON data source or configure the connection. Just follow the three steps below.

5.1. Log in to Redash

Open http://localhost:5000/ and log in with the default credentials:

5.2. Change the default login and password

For security reasons change the default email and password immediately: in Redash click Settings (bottom-left corner) → Account and set your own credentials.

5.3. Generate a token and paste it into OmniMES

1. In Redash go to Settings → Account.

2. In the API Key section click Regenerate to generate a new token, then copy it.

3. In OmniMES go to Settings → Redash and fill in:

   • Redash address: localhost:5000

   • API Key: the freshly generated token you copied

From now on OmniMES is connected to Redash - you can use the ready-made reports and dashboards.

6. Ollama configuration (AI assistant)

The Ollama server provides local language models for the built-in AI assistant in OmniMES. It is installed automatically together with the full version of the system (the install_all_* scripts).

6.1. Hardware requirements

The default installation uses the CPU, so performance depends on the processor's power:

• CPU (minimum): a 24-core desktop processor or stronger.

• GPU (recommended): an NVIDIA 3070 series card or higher (one or more, depending on the chosen AI model).

💡 Recommendation: For larger deployments, run Ollama on a separate server (Docker or natively).

6.2. Installation on a separate server

• Docker image: https://hub.docker.com/r/ollama/ollama

• Linux: https://github.com/ollama/ollama/blob/main/docs/linux.md

• Windows: https://github.com/ollama/ollama/blob/main/docs/windows.md

docker-compose.yaml

version: '3.8'

services:
  ollama:
    image: ollama/ollama
    environment:
      - OLLAMA_ORIGINS="*"
      - OLLAMA_HOST="0.0.0.0:11434"
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama:/root/.ollama
    restart: unless-stopped

volumes:
  ollama:

For external access: host = 0.0.0.0, port = 11434, CORS = "*" (or narrowed down as needed).

6.3. Downloading an AI model

List of supported models: https://ollama.com/library

Downloading the chosen model:

ollama pull llama3

6.4. Linking with OmniMES

In OmniMES Settings:

1. Select the AI type: internal.

2. Enter the Ollama server address (e.g. http://192.168.3.100:11434).

3. The AI model version field should show the list of downloaded models.

If the connection fails, an error message appears in the field - check the server address and the open ports.

7. Changing system passwords

The PostgreSQL database or MQTT broker password is changed with the included **set_pass.py** script. The script operates on the running containers (docker compose exec database psql), so the OmniMES stack must be running, and the script itself only requires an installed Python and a working Docker - no extra libraries (no host, port, or old password needed). Run it in the directory containing the OmniMES stack docker-compose.yaml.

7.1. Changing the MQTT broker password (user emqx)

# Windows
py set_pass.py --newpassbroker NEW_BROKER_PASSWORD
# Linux / macOS
python3 set_pass.py --newpassbroker NEW_BROKER_PASSWORD

The script updates the MQTT emqx user password in the database, the stream .env file, and restarts the stream service.

7.2. Changing the PostgreSQL database password (user omni)

# Windows
py set_pass.py --newpassdb NEW_DATABASE_PASSWORD
# Linux / macOS
python3 set_pass.py --newpassdb NEW_DATABASE_PASSWORD

The script changes the omni user password, updates omnimes/.env, the EMQX auth configuration (emqx_auth_pgsql.conf), the API config.json and the stream .env, then restarts the broker, api and stream services - nothing has to be done manually.

8. System update

Each main component can be updated independently using dedicated scripts:

Each script performs the sequence: pull → rebuild → restart the container.

💡 Linux: remember the permissions - chmod +x updated_*.sh before the first run.

8.1. Full environment rebuild

In case of problems (a damaged container, configuration inconsistency) you can force a rebuild:

• OmniMES: recreated_omnimes.bat (Windows)

• Redash: recreated_redash.bat (Windows)

⚠️ The rebuild does not remove the postgres_data volume with the database data, but a backup is recommended before running it.

9. Uninstalling

9.1. Procedure

In the installation folder, run the script:

• Windows: uninstall.bat

• Linux / macOS: uninstall.sh

On Linux, remember to grant permissions first:

chmod +x uninstall.sh
./uninstall.sh

9.2. Warning - data loss

⚠️ Uninstalling removes all containers, including the **database** container (PostgreSQL + TimescaleDB), which stores the data from the production devices.

Before uninstalling, back up the database - via Adminer (http://localhost:8079) or with docker compose exec database pg_dump -U omni -Fc -d data > data.dump (similarly for the omni database). Restore: pg_restore.

10. Exposing OmniMES outside the computer

10.1. What to do when port 80 is busy?

By default the OmniMES web interface runs on port 80. If the port is busy:

• Option A: free port 80 in the operating system (stop the service that uses it).

• Option B: change the port mapping in docker-compose.yaml - in the www section change the left number from 80:80 to e.g. 8080:80. After this, OmniMES will be available at http://<address>:8080.

10.2. Exposing outside the local network

The configuration must be done on the router, directing traffic to the IP address of the computer where the system is installed. The required port forwarding:

10.3. Connecting an external device (HMI / gateway)

On the device (HMI panel, communication gateway, etc.) specify:

• the OmniMES server address, e.g. 192.168.3.100

• the MQTT port: 1883 (standard)

• the EMQX broker login data:

  • Login: emqx

  • Password: Fb7jCkux (factory - change it after deployment, see section 7)

10.4. Windows as a server (not recommended) - automatic Docker startup

If the system has to run on Windows anyway, enable the "Start Docker Desktop when you log in" option in Docker Desktop. Keep in mind that Docker Desktop runs in the session of a logged-in user - after a restart (e.g. after Windows updates) OmniMES starts only once a user logs in.

11. Installation files - reference

11.1. Full installation

11.2. Modular installation

11.3. Single-component updates

11.4. Environment rebuild

11.5. Uninstalling

11.6. Tools

12. .env configuration files

The installation package contains two **.env** files that configure the system. Set them before the first run of the installation script.

12.1. omnimes/.env - OmniMES stack

12.2. redash/.env - Redash stack

✉️ Redash SMTP mail - set it BEFORE installation (if needed). The REDASH_MAIL_SERVER / REDASH_MAIL_PORT / REDASH_MAIL_USE_TLS / REDASH_MAIL_USE_SSL / REDASH_MAIL_USERNAME / REDASH_MAIL_PASSWORD / REDASH_MAIL_DEFAULT_SENDER fields are needed only for email alert notifications, user invitations and password reset. Redash works without them.

• If you want emails → fill these fields with your own SMTP server before installation.

• If you do not use email → these lines must stay commented out. Never leave an empty line like REDASH_MAIL_PORT= - Redash runs int('') and the container goes into a restart loop.

• After editing .env, recreate the container (docker compose up -d), not just restart - Docker Compose reads .env when creating the container, not on restart.

13. Moving data to another server (database dumps)

The system can be moved with its data to another server. On the old server, create dumps from the database container, copy them to omnimes/database/base/ before the build, and the installation script will restore everything on the new server.

docker compose exec database pg_dump -U omni -Fc -d omni        > omni.dump        # configuration
docker compose exec database pg_dump -U omni -Fc -d omni_queue  > omni_queue.dump  # Django Q queue
docker compose exec database pg_dump -U omni -Fc -d mqtt        > mqtt.dump        # EMQX auth
docker compose exec database pg_dump -U omni -Fc -d redash      > redash.dump      # dashboards/queries/users
docker compose exec database pg_dump -U omni -Fc -d data        > data.dump        # telemetry (Timescale hypertable)

Copy the selected files to instalka3/omnimes/database/base/ and run the installation.

ℹ️ No dumps = a fresh install: the schema is created by migrate, and the data database is fed by stream. Details and accepted formats (.dump / .sql / .sql.gz) are described in database/base/README.md.