Integrated Ollama and Open WebUI on ubuntu 24 LTS hardened by Hanwei

This product provides an integrated local LLM environment on Ubuntu 24.04.4 LTS x86_64. The delivered configuration combines Ollama 0.34.0, Open WebUI 0.11.3, Docker Engine 29.8.0, NVIDIA Driver 580.173.02, and a preloaded Qwen2.5 7B model. It configures Open WebUI for the Ollama service and uses a Docker volume for WebUI application data. The additional charges cover seller-performed runtime and web-interface integration, model preparation, local API and WebUI data-volume configuration, and testing on g4dn.xlarge with NVIDIA Tesla T4 16 GB. Support and maintenance are supplementary services and are not the sole basis for the software charges.

What the additional charges cover:

• Integration of Ollama 0.34.0 and Open WebUI 0.11.3 into one local browser and API model-serving environment.
• Preparation of qwen2.5:7b in the Ollama model environment for access through the runtime and WebUI.
• Configuration and documentation of the Ollama API on TCP 11434, the Open WebUI interface on TCP 8080, and the WebUI application-data volume.
• Testing on g4dn.xlarge with NVIDIA Tesla T4 16 GB covering GPU detection, service and container state, model inventory, local API inference, restart-policy configuration, and volume mapping.
• Operational guidance for component verification, initial administrator setup, API access boundaries, backups, and customer-controlled network exposure.

Technical implementation:

• Ubuntu 24.04.4 LTS x86_64 provides the operating-system baseline.
• NVIDIA Driver 580.173.02 provides the recorded GPU driver layer for the tested Tesla T4 configuration.
• Ollama 0.34.0 provides the local model runtime and API on TCP 11434 and is managed by ollama.service.
• Open WebUI application 0.11.3 runs in the open-webui Docker container and provides the browser interface on TCP 8080.
• Docker Engine 29.8.0 provides the container runtime.
• The open-webui Docker volume is mapped to /app/backend/data for WebUI application data.
• qwen2.5:7b is the model name used in the Ollama environment.

Included components:

• Ubuntu 24.04.4 LTS x86_64.
• NVIDIA Driver 580.173.02.
• Ollama 0.34.0.
• Open WebUI application 0.11.3.
• Docker Engine 29.8.0.
• Preloaded Qwen2.5 7B model as qwen2.5:7b.
• Product-specific usage, initial-access, and network-boundary documentation.

Tested deployment baseline:

• Instance type: g4dn.xlarge.
• GPU: NVIDIA Tesla T4 with 16 GB VRAM.
• Operating system: Ubuntu 24.04.4 LTS x86_64.
• Functional checks: GPU detection, Ollama service state, Docker and Open WebUI container state, model inventory, local API inference, Docker restart-policy configuration, and WebUI data-volume mapping.
• Scope boundary: other EC2 instances, GPUs, CPU-only and multi-GPU configurations, performance levels, concurrency, latency, throughput, high availability, and recovery behavior are outside this tested baseline.

Security and operational boundaries:

• SSH access uses the EC2 key pair and should be restricted to trusted administrative CIDRs.
• Open WebUI uses TCP 8080. Limit access to trusted networks, complete initial administrator setup, and configure customer-managed TLS or an authenticated reverse proxy before broader access.
• Ollama uses TCP 11434 and does not provide application-layer API authentication in the recorded configuration. Keep it limited to localhost or trusted VPC clients, or place it behind a customer-managed authenticated proxy and TLS.
• Verify actual listening addresses and the Open WebUI-to-Ollama connection before enabling remote access.
• Updates, model retrieval, and optional WebUI capabilities may access external services. Customers control outbound access and are responsible for applicable third-party terms.
• Customers are responsible for Security Groups, VPC design, identity controls, TLS, operating-system and application updates, backups, monitoring, request and response data, model suitability, model licensing, and applicable legal requirements.
• No certification, upstream endorsement, performance guarantee, broad hardware compatibility, recovery guarantee, or data-residency guarantee is claimed.

Suitable use cases:

• Customer-controlled browser interaction with the preloaded Qwen2.5 7B model through Open WebUI.
• Internal application integration with the local Ollama API inside a customer-managed VPC.
• GPU-backed model-serving workloads using the tested g4dn.xlarge/Tesla T4 software baseline with customer sizing and backup policies.

Pricing statement:

• Marketplace software charges cover the integrated software configuration and tested deployment baseline described above. EC2, EBS, data transfer, and other AWS infrastructure charges are billed separately. External repositories or third-party services selected by the customer may have separate terms or charges.