fix(context): resolve llama-server host port from OLLAMA_PORT and default to 11434

[vaibhavsrv]

Why this matters

In ods/scripts/build-installation-context.py, the builder constructs the dynamic 'About this installation' context block by querying _loaded_model(llm_port=...) against the local host. However, the host port resolution logic only inspected LLM_PORT and LLAMACPP_PORT, omitting OLLAMA_PORT and LLAMA_SERVER_PORT, and defaulted to 8080. In ODS Compose topologies, 8080 is the container-internal port while the host publishes llama-server on ${OLLAMA_PORT:-11434}. As a result, when run on default installations or where OLLAMA_PORT is customized, the script probed 127.0.0.1:8080 where nothing was bound on the host, silently failing to detect the active model. Additionally, invoking _loaded_model with keyword arguments broke 0-argument test mocks.

This patch updates the environment port resolution in build_context_block to prioritize OLLAMA_PORT and LLAMA_SERVER_PORT before falling back to LLM_PORT and defaulting to 11434, updates the default parameter of _loaded_model to 11434, and wraps the call to handle 0-argument mock callables cleanly.

Validation

• Repro baseline: Evaluated default environment block generation without LLM_PORT and with OLLAMA_PORT=11435; confirmed _loaded_model was called with port 8080 and test_persona_running_containers.py failed with TypeError.
• Post-fix: Verified all 3 unit tests in test_build_installation_context_port.py and 4 parameterized tests in test_persona_running_containers.py pass with exit code 0.
• Context suites: 7 passed. New-test Ruff and diff checks pass; new regressions wired into Linux CI.

Overlap check

• fix(env): probe .env keys literally in _env_set and mode-switch env_set #5819 (0xacee): Addressed macOS environment generation and safe escaping; touches ods/installers/macos/ with zero overlap in context builders.
• fix(cold-storage): create cold directory before archiving idle models #5861 (vaibhavsrv): Addressed directory creation in cold storage archiving; touches ods/scripts/llm-cold-storage.sh.
• fix(scripts): validate context length, port bounds, and control characters in render-runtime-configs #5056 (upstream): Touched Compose model definitions without altering host context inspection.

Risk / AI disclosure

AI-assisted investigation, implementation and CLI regressions. This strengthens host port resolution and mock interoperability for local installation context generation, not runtime model routing. Independent human review and platform/runtime qualification remain gates. No running configuration, deployment or upstream merge changed.

Follow-up integration evidence

Composed with #5861 at 2a24f49 without conflicts. Production and test diffs passed together; persona running container contracts remain intact.
Backlog composition was local-only (production/test diffs, excluding workflow/Makefile wiring); it is not an upstream merge or independent human approval. Declared live-review gates remain open.

[Lightheartdevs]

Thanks for this contribution. public-beta was promoted into main on 2026-09-24 and no longer receives changes, so we're closing pull requests that target it. This isn't a judgment on the change itself. If it's still needed, please rebase onto main and open a focused PR. See #7253 for details and the contribution policy.