Skip to content

dashboard: rename Ollama endpoints to /api/llm, remove dead registry browse (PR2)#66

Closed
AlienWalker1995 wants to merge 1 commit into
chore/remove-ollama-refs-pr1from
chore/rename-ollama-endpoints-pr2
Closed

dashboard: rename Ollama endpoints to /api/llm, remove dead registry browse (PR2)#66
AlienWalker1995 wants to merge 1 commit into
chore/remove-ollama-refs-pr1from
chore/rename-ollama-endpoints-pr2

Conversation

@AlienWalker1995

Copy link
Copy Markdown
Owner

Stacked on #65

Base is the PR1 branch (chore/remove-ollama-refs-pr1), so this shows only the delta. Merge #65 first; GitHub will retarget this to main. Together they leave zero Ollama references in the repo (except CHANGELOG.md history and Open WebUI's ENABLE_OLLAMA_API=false off-switch).

What

The dashboard's /api/ollama/* endpoints were actually managing GGUF/llama.cpp models (misnamed). This renames them to a clean /api/llm/* namespace and removes the one genuinely-dead piece.

  • dashboard/app.py: /api/ollama/{models,delete,unload,pull,pull/status,ps}/api/llm/*; delete the dead /api/ollama/library + _fetch_ollama_library + registry-library config vars; rename _run_ollama_pull_run_gguf_pull, _ollama_pull_status_gguf_pull_status, _hf_url_to_ollama_hf_url_to_repo; reword user-facing strings off Ollama. Behavior-preserving for the 6 live endpoints.
  • dashboard/static/index.html: retarget the 5 UI fetch sites; remove the "browse Ollama registry" dropdown (kept pull-by-Hugging-Face-repo input); rename all Ollama UI labels/identifiers. (/api/llm/* chosen because /api/models/* is already ComfyUI's.)
  • tests: drop the library test; repoint the global-exception-handler test to /api/llm/models via a patched _scan_gguf_models.
  • docker-compose.yml: open-webui OLLAMA_BASE_URL=ENABLE_OLLAMA_API=false (explicit disable); drop n8n OLLAMA_HOST (it pointed at the LiteLLM gateway, which doesn't speak Ollama's API; n8n's real path is OPENAI_API_BASE_URL).
  • Docs: README "Ollama models" section + component-dashboard-ui.md updated to /api/llm.

Validation

  • python -m pytest tests/394 passed (the 2 failures are the known local-only hermes_socket/secrets_isolation tests — they pass/skip in CI).
  • UI↔backend route cross-check: all 5 UI fetches map to real routes (/api/llm/unload has no UI caller, matching the pre-existing state).
  • git grep -in ollama across the combined stack → only CHANGELOG.md + ENABLE_OLLAMA_API.
  • index.html re-normalized to its base CRLF so the diff is real edits only (313 lines), not a whole-file EOL flip.

🤖 Generated with Claude Code

…browse

The dashboard's /api/ollama/* endpoints were actually managing GGUF/llama.cpp
models (misnamed). Rename them to a clean /api/llm/* namespace and remove the
one genuinely-dead piece (the Ollama community-registry browse).

- app.py: /api/ollama/{models,delete,unload,pull,pull/status,ps} -> /api/llm/*;
  delete /api/ollama/library + _fetch_ollama_library + library config vars;
  rename _run_ollama_pull->_run_gguf_pull, _ollama_pull_status->_gguf_pull_status,
  _hf_url_to_ollama->_hf_url_to_repo; reword user-facing strings off Ollama
- index.html: retarget fetches to /api/llm/*, remove the registry-browse dropdown
  (kept pull-by-HF-repo input), rename all Ollama UI labels/identifiers
- tests: drop the library test; repoint the global-exception-handler test to
  /api/llm/models via a patched _scan_gguf_models
- docker-compose.yml: open-webui OLLAMA_BASE_URL= -> ENABLE_OLLAMA_API=false;
  drop n8n OLLAMA_HOST (pointed at the LiteLLM gateway, which is not an Ollama API)
- README "Ollama models" section + component-dashboard-ui.md updated to /api/llm

Behavior-preserving for the 6 live endpoints. pytest: 394 passed (2 pre-existing
local-only failures unrelated). Stacked on the PR1 doc/script scrub.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@AlienWalker1995 AlienWalker1995 deleted the branch chore/remove-ollama-refs-pr1 July 1, 2026 20:08
@AlienWalker1995 AlienWalker1995 deleted the chore/rename-ollama-endpoints-pr2 branch July 1, 2026 20:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant