litellm/tests/test_litellm/integrations/otel
Yassin Kortam 440b1bcf65
fix(otel): make OTLP export work against Grafana Cloud (#35060)
Three defects kept LiteLLM's OTel metrics from reaching an OTLP backend.

OTEL_EXPORTER_OTLP_HEADERS is W3C Baggage encoded per the OTLP spec, so its
values are percent-encoded. litellm split the string on "," and "=" and passed
the raw value straight to the exporter, so a vendor that documents
"Authorization=Basic%20<token>" got a literal "%20" on the wire and the backend
rejected the credential. Grafana Cloud documents exactly that shape, which made
its OTLP gateway unreachable. Header parsing now delegates to the OTel SDK's own
W3C Baggage parser in liberal mode, so percent-encoded values decode and values
that were never encoded keep working. It moves from model/utils.py to
plumbing/providers.py because model/ is deliberately free of opentelemetry
imports; providers.parse_headers was already the entry point every caller used.

The OTLP metric exporters then overrode histogram temporality to delta.
Prometheus and Mimir, which back Grafana Cloud's OTLP gateway, reject delta
histograms outright: the gateway answers 400 "invalid temporality and type
combination" and drops the entire batch, so every GenAI metric was silently lost
while traces kept flowing. Backends that prefer delta still accept cumulative, so
the SDK default is the compatible choice in both directions, and the enterprise
billing exporter already relies on it.

Three GenAI instruments also carried names no convention or backend defines, so
nothing downstream could chart them. Time to first token and time per output
token take their semconv names, gen_ai.server.time_to_first_token and
gen_ai.server.time_per_output_token; the gen_ai.client.response.* spellings
litellm used are not conventions at all. Cost has no semconv instrument, so it
takes gen_ai.usage.cost, the name backends already query for spend. All three are
listed verbatim in Grafana Cloud's AI Observability integration reference, so its
prebuilt panels find them. Both engines now read the names from the shared Metric
constants rather than repeating string literals, so v1 and v2 cannot drift.

The renames are breaking for anyone charting the former names; the docs and the
release changelog carry the migration note.
2026-07-29 13:43:33 -07:00
..
test_otel_v2_baggage.py feat(otel): allowlist team_metadata sub-keys promoted to baggage (#29442) 2026-06-01 14:02:23 -07:00
test_otel_v2_components.py fix(otel): make OTLP export work against Grafana Cloud (#35060) 2026-07-29 13:43:33 -07:00
test_otel_v2_config_baggage_parenting_guardrails.py feat(otel): typed semconv-aligned OpenTelemetry instrumentation (#28909) 2026-05-29 23:15:27 -07:00
test_otel_v2_dynamic.py fix(otel): one v2 logger owns the global provider; scope tenant OTLP creds per exporter (#30590) 2026-06-19 11:15:29 -07:00
test_otel_v2_emitter.py fix(otel): restore proxy-level error.* attributes on v2 failure spans (LIT-4179) (#33664) 2026-07-18 10:52:27 -07:00
test_otel_v2_logger.py fix(otel): make OTLP export work against Grafana Cloud (#35060) 2026-07-29 13:43:33 -07:00
test_otel_v2_metrics.py fix(otel): make OTLP export work against Grafana Cloud (#35060) 2026-07-29 13:43:33 -07:00
test_otel_v2_mount.py perf(otel): resolve LITELLM_OTEL_V2 flag once instead of rebuilding settings per call (#30989) 2026-06-22 11:26:42 -07:00
test_otel_v2_multibackend.py feat(otel): typed semconv-aligned OpenTelemetry instrumentation (#28909) 2026-05-29 23:15:27 -07:00
test_otel_v2_presets.py fix(otel): point AgentOps OTLP exporter at otlp.agentops.ai (#31490) 2026-06-26 20:39:39 -07:00
test_otel_v2_sources_of_truth.py refactor(otel): move litellm error detail keys under the litellm.* namespace (#32591) 2026-07-09 00:51:37 -07:00
test_otel_v2_vendor_mappers.py feat(otel): typed semconv-aligned OpenTelemetry instrumentation (#28909) 2026-05-29 23:15:27 -07:00
test_runtime.py perf(otel): memoize per-request lazy import of otel runtime hooks (#31707) 2026-06-30 10:26:20 -07:00