Skip to content

test(litellm): pin region and rerun the cache test #5624

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zastrowm merged 1 commit into from
Oct 28, 2026
Merged

test(litellm): pin region and rerun the cache test #5624

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 16 additions & 4 deletions strands-py/tests_integ/models/test_model_litellm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,20 +10,29 @@
from strands.models.litellm import LiteLLMModel
from tests_integ.conftest import retry_on_flaky

_MODEL_ID = "bedrock/global.anthropic.claude-sonnet-4-6"

# Pinned so an explicitly-exported AWS_REGION cannot redirect these tests: conftest only
# setdefault()s the region, so a developer's own AWS_REGION still wins. Bedrock prompt-cache
# read-back measures ~88% in us-west-2 against ~100% in us-east-1 and us-east-2, so the region
# decides whether the cache assertions below are stable. us-east-2 is the known-good alternative.
_REGION = "us-east-1"
_CLIENT_ARGS = {"aws_region_name": _REGION}


@pytest.fixture
def model():
return LiteLLMModel(model_id="bedrock/global.anthropic.claude-sonnet-4-6")
return LiteLLMModel(model_id=_MODEL_ID, client_args=_CLIENT_ARGS)


@pytest.fixture
def streaming_model():
return LiteLLMModel(model_id="bedrock/global.anthropic.claude-sonnet-4-6", params={"stream": True})
return LiteLLMModel(model_id=_MODEL_ID, client_args=_CLIENT_ARGS, params={"stream": True})


@pytest.fixture
def non_streaming_model():
return LiteLLMModel(model_id="bedrock/global.anthropic.claude-sonnet-4-6", params={"stream": False})
return LiteLLMModel(model_id=_MODEL_ID, client_args=_CLIENT_ARGS, params={"stream": False})


@pytest.fixture
Expand Down Expand Up @@ -259,7 +268,10 @@ def test_streaming_returns_usage_metrics(model_fixture, request):
assert result.metrics.accumulated_usage["totalTokens"] > 0


@pytest.mark.skip(reason="Turn-2 prompt-cache read intermittently returns 0 tokens; see #5315")
# Bedrock prompt-cache read-back is best effort: an entry written on turn 1 is occasionally not
# readable on turn 2, which the provider reports as a fresh write of identical size. Reruns rather
# than `retry_on_flaky`, whose wrapper is sync and breaks async test collection. See #5315.
@pytest.mark.flaky(reruns=4, reruns_delay=1)
@pytest.mark.asyncio
@pytest.mark.parametrize("streaming", [True, False])
async def test_cache_read_tokens_multi_turn(model, streaming):
Expand Down
Loading