bug: LANGFUSE_MAX_EVENT_SIZE_BYTES is parsed but never enforced for score events
Maintainer antworten meist innerhalb von 1 Tag
Bewertung
Dieses Issue wurde noch nicht bewertet.
Beschreibung
Describe the bug
MAX_EVENT_SIZE_BYTES = int(os.environ.get("LANGFUSE_MAX_EVENT_SIZE_BYTES", 1_000_000)) (score_ingestion_consumer.py#L18) is not referenced anywhere else in the repo (grep -rn MAX_EVENT_SIZE langfuse only finds the definition). The env var was added in #1103 ("make max event and batch size configurable"), but it currently has no effect: an oversized score event is posted as-is instead of being dropped or truncated.
Secondary point: in _next() the running total is compared to MAX_BATCH_SIZE_BYTES only after the event has already been appended (score_ingestion_consumer.py#L83-L108), so a batch can exceed the batch limit by up to one event (and since single events are not capped, by an arbitrary amount).
Expected: events above LANGFUSE_MAX_EVENT_SIZE_BYTES are dropped (with a logged error) or truncated, and the batch limit is checked before appending.
Actual: a 10 MB score comment is sent in a single request. Whether the server then rejects it and how that is surfaced depends on the 4xx handling; today a rejected batch is dropped without a log line (see #1874, which is about that part). Users who set the env var, as in langfuse/langfuse#7027, get no client-side effect from it.
Severity is low to medium: it only affects unusually large score payloads, but the env var suggests protection that is not there.
Steps to reproduce
import httpx
from queue import Queue
from langfuse._utils.request import LangfuseClient
from langfuse._task_manager.score_ingestion_consumer import ScoreIngestionConsumer, MAX_EVENT_SIZE_BYTES, MAX_BATCH_SIZE_BYTES
sizes = []
def h(req): sizes.append(len(req.content)); return httpx.Response(200, json={})
cl = LangfuseClient("pk", "sk", "http://x", "1", 5, httpx.Client(transport=httpx.MockTransport(h)))
q = Queue(); c = ScoreIngestionConsumer(ingestion_queue=q, identifier=0, client=cl, public_key="pk", flush_at=15)
q.put({"id": "big", "type": "score-create", "body": {"comment": "x" * 10_000_000}})
c.upload()
print("MAX_EVENT", MAX_EVENT_SIZE_BYTES, "MAX_BATCH", MAX_BATCH_SIZE_BYTES, "posted bytes", sizes)
Observed output:
MAX_EVENT 1000000 MAX_BATCH 2500000 posted bytes [10000173]
Langfuse Cloud or self-hosted?
Either; the behaviour is client side only.
If self-hosted, what version are you running?
n/a
SDK and integration versions
langfuse 4.17.0 (main @ bf11ec121479145923ff27b32ec2ad1d06fd2d51), Python 3.14.0.
Additional information
Possible direction: after computing item_size in _next(), if item_size > MAX_EVENT_SIZE_BYTES: logger.error(...); self._ingestion_queue.task_done(); continue, and check total_size + item_size > MAX_BATCH_SIZE_BYTES before appending, carrying the event into the next batch. Alternatively remove the constant and the env var if client-side enforcement is not wanted.
Minor and separate: int(os.environ[...]) at import time makes import langfuse raise on a malformed value (this was also noted in the review of #1103).
Related: #1874 (4xx batch drops are not logged), langfuse/langfuse#7027 (413 from the server with LANGFUSE_MAX_EVENT_SIZE_BYTES set).
Are you interested in contributing a fix for this bug?
Yes
- Vorherrschende Sprache
- Python
- Sterne
- 498
- Forks
- 361
- Ø Merge
- 16 Std. 40 Min.
- Gemergte PRs (30 T.)
- 38
Entwicklungsumgebung
- Kein Dockerfile und keine Docker-Compose-Datei
- Hat eine Pull-Request-Vorlage
- Beitragsleitfaden lesen
Erste Schritte
- Lesen Sie das ganze Issue und danach den Beitragsleitfaden des Projekts.
- Schreiben Sie ins Issue, dass Sie es übernehmen — das erspart doppelte Arbeit.
- Forken Sie das Repository und arbeiten Sie in einem Branch.
- Öffnen Sie einen Pull Request, der die Issue-Nummer nennt.
Mehr aus langfuse/langfuse-python
-
tests/unit/test_prompt.py fails on a fresh clone without LANGFUSE_* env varsEvtl. vergeben @Mathoholic hat das heute übernommen. Offenbug feat-prompt-management sdk-python
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 80/100
langfuse/langfuse-python#1976 · 3 Kommentare ·
Maintainer antworten meist innerhalb von 1 Tag
-
bug(langchain): failed streaming LLM runs retain IDs in callback memoEvtl. vergeben @hassiebp hat das vor 1 Tag übernommen. Offenbug integration-langchain sdk-python
langfuse/langfuse-python#1977 · 1 zugewiesene Person ·
Maintainer antworten meist innerhalb von 1 Tag
-
bug: ChatPromptClient.compile appends str(whole list) once per non-dict placeholder itemEvtl. vergeben @hassiebp hat das vor 2 Tagen übernommen. Offenbug feat-prompt-management sdk-python
langfuse/langfuse-python#1971 · 1 zugewiesene Person ·
Maintainer antworten meist innerhalb von 1 Tag
-
bug: non-numeric status in a 207 error entry raises inside handle_exception, kills the score consumer thread, flush()/shutdown() then hangEvtl. vergeben @hassiebp hat das vor 2 Tagen übernommen. Offen[Integrations] Language Clients bug feat-scores sdk-python
langfuse/langfuse-python#1967 · 1 zugewiesene Person ·
Maintainer antworten meist innerhalb von 1 Tag
-
bug: Partial-failure (207) ingestion response retries the whole batch up to 3x; per-item error details only logged at DEBUGEvtl. vergeben @hassiebp hat das vor 2 Tagen übernommen. Offenbug feat-ingestion
langfuse/langfuse-python#1965 · 1 zugewiesene Person ·
Maintainer antworten meist innerhalb von 1 Tag
Alle Issues in langfuse/langfuse-python
Ähnliche Issues
-
bug
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 76/100
lugassawan/swe-workbench#799 ·
Maintainer antworten meist innerhalb von 1 Tag
-
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 74/100
vllm-project/vllm#61131 ·
Maintainer antworten meist innerhalb von 1 Tag
-
Non-string form field names raise AttributeError when files are attachedEvtl. vergeben @nightcityblade hat das heute übernommen. Offen
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 82/100
-
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 78/100
kubernetes-client/python#2735 ·
Maintainer antworten meist innerhalb von 1 Tag
-
desktop help wanted test
Schwierigkeit 2/5 1-3 Stunden Anfängerfreundlichkeit 78/100
Maintainer antworten meist innerhalb von 1 Tag