Compare commits

..

1 Commits

Author SHA1 Message Date
srikanthccv
487d00b6c9 test: add semantic convention phase one closure gate 2026-08-07 01:20:55 +05:30
2 changed files with 243 additions and 0 deletions

View File

@@ -220,6 +220,10 @@ py-test-teardown: ## Tear down the shared SigNoz backend
py-test: ## Runs integration tests
@cd tests && uv run pytest --basetemp=./tmp/ -vv --capture=no integration/tests/
.PHONY: py-test-semconv-phase1
py-test-semconv-phase1: py-test-setup ## Rebuild the shared stack and run the semantic-convention Phase 1 matrix
@cd tests && uv run pytest --basetemp=./tmp/ -vv --reuse --capture=no integration/tests/queriertraces/13_semconv_evolution.py
.PHONY: py-clean
py-clean: ## Clear all pycache and pytest cache from tests directory recursively
@echo ">> cleaning python cache files from tests directory"

View File

@@ -0,0 +1,239 @@
"""Phase 1 end-to-end checks for semantic-convention name evolution.
The fixture models a fleet split across SDK generations and deliberately includes
a dual-emitting conflict. Both request spellings must address one logical field,
with the current spelling winning when a row contains both.
"""
from collections.abc import Callable, Generator
from datetime import UTC, datetime, timedelta
from http import HTTPStatus
from typing import Any
import pytest
import requests
from fixtures import types
from fixtures.auth import USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD
from fixtures.querier import Aggregation, BuilderQuery, OrderBy, RequestType, TelemetryFieldKey, make_query_request
from fixtures.traces import TraceIdGenerator, Traces, TracesKind, TracesStatusCode
CURRENT = "deployment.environment.name"
OLD = "deployment.environment"
PREFIX = "semconv-phase1"
PRODUCTION_SPANS = {
f"{PREFIX}-old",
f"{PREFIX}-current",
f"{PREFIX}-both",
f"{PREFIX}-conflict",
}
STAGING_SPANS = {f"{PREFIX}-staging"}
MISSING_SPANS = {f"{PREFIX}-missing"}
def _span(timestamp: datetime, suffix: str, environment: dict[str, str]) -> Traces:
service = f"{PREFIX}-{suffix}"
return Traces(
timestamp=timestamp,
duration=timedelta(milliseconds=10),
trace_id=TraceIdGenerator.trace_id(),
span_id=TraceIdGenerator.span_id(),
name=service,
kind=TracesKind.SPAN_KIND_SERVER,
status_code=TracesStatusCode.STATUS_CODE_OK,
resources={"service.name": service, **environment},
attributes=dict(environment),
)
@pytest.fixture(name="semconv_phase1_data")
def semconv_phase1_data(
insert_traces: Callable[[list[Traces]], None],
clickhouse: types.TestContainerClickhouse,
) -> Generator[datetime]:
now = datetime.now(tz=UTC).replace(microsecond=0) - timedelta(minutes=2)
insert_traces(
[
_span(now - timedelta(seconds=5), "old", {OLD: "production"}),
_span(now - timedelta(seconds=4), "current", {CURRENT: "production"}),
_span(now - timedelta(seconds=3), "both", {OLD: "production", CURRENT: "production"}),
_span(now - timedelta(seconds=2), "conflict", {OLD: "staging", CURRENT: "production"}),
_span(now - timedelta(seconds=1), "staging", {OLD: "staging"}),
_span(now, "missing", {}),
]
)
# Service-map rows are derived by the collector in production. Seed the
# derived table directly here so the backend alias allowlist is tested in
# isolation; the collector repository owns its write-path integration test.
for environment, suffix in (("production", "production"), ("staging", "staging")):
clickhouse.conn.command(
f"""
INSERT INTO signoz_traces.distributed_dependency_graph_minutes_v2
(src, dest, duration_quantiles_state, error_count, total_count, timestamp,
deployment_environment, k8s_cluster_name, k8s_namespace_name)
SELECT
'{PREFIX}-map-{suffix}', '{PREFIX}-map-child',
quantilesState(0.5, 0.75, 0.9, 0.95, 0.99)(toFloat64(1000000)),
toUInt64(0), toUInt64(1), toDateTime({int(now.timestamp())}),
'{environment}', '', ''
"""
)
yield now
cluster = clickhouse.env["SIGNOZ_TELEMETRYSTORE_CLICKHOUSE_CLUSTER"]
clickhouse.conn.command(
f"ALTER TABLE signoz_traces.dependency_graph_minutes_v2 ON CLUSTER '{cluster}' "
f"DELETE WHERE startsWith(src, '{PREFIX}-map-') SETTINGS mutations_sync = 1"
)
def _result(response: requests.Response) -> dict[str, Any]:
assert response.status_code == HTTPStatus.OK, response.text
results = response.json()["data"]["data"]["results"]
assert len(results) == 1
return results[0]
def _raw_names(
signoz: types.SigNoz,
token: str,
now: datetime,
expression: str,
) -> set[str]:
response = make_query_request(
signoz,
token,
start_ms=int((now - timedelta(minutes=2)).timestamp() * 1000),
end_ms=int((now + timedelta(minutes=1)).timestamp() * 1000),
request_type=RequestType.RAW,
queries=[
BuilderQuery(
signal="traces",
name="A",
limit=100,
filter_expression=expression,
select_fields=[TelemetryFieldKey("span.name")],
order=[OrderBy(TelemetryFieldKey("timestamp"), "asc")],
).to_dict()
],
)
return {row["data"]["name"] for row in (_result(response).get("rows") or [])}
def _metadata_values(signoz: types.SigNoz, token: str, name: str, context: str) -> set[str]:
response = requests.get(
signoz.self.host_configs["8080"].get("/api/v1/fields/values"),
timeout=5,
headers={"authorization": f"Bearer {token}"},
params={
"signal": "traces",
"name": name,
"fieldContext": context,
"fieldDataType": "string",
},
)
assert response.status_code == HTTPStatus.OK, response.text
return set(response.json()["data"]["values"].get("stringValues") or [])
def test_semconv_phase1_mixed_sdk_generations( # pylint: disable=too-many-statements
signoz: types.SigNoz,
create_user_admin: None, # pylint: disable=unused-argument
get_token: Callable[[str, str], str],
semconv_phase1_data: datetime,
) -> None:
now = semconv_phase1_data
token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
# Resource and span-attribute paths share the same matrix. Run every
# operator with both the saved-query (old) and current request spellings.
for context in ("resource", "attribute"):
for requested in (CURRENT, OLD):
field = f"{context}.{requested}"
assert _raw_names(signoz, token, now, f"{field} = 'production'") == PRODUCTION_SPANS
assert _raw_names(signoz, token, now, f"{field} = 'staging'") == STAGING_SPANS
assert _raw_names(signoz, token, now, f"{field} != 'production'") == STAGING_SPANS
assert _raw_names(signoz, token, now, f"{field} EXISTS") == PRODUCTION_SPANS | STAGING_SPANS
assert _raw_names(signoz, token, now, f"{field} NOT EXISTS") == MISSING_SPANS
grouped = make_query_request(
signoz,
token,
start_ms=int((now - timedelta(minutes=2)).timestamp() * 1000),
end_ms=int((now + timedelta(minutes=1)).timestamp() * 1000),
request_type=RequestType.SCALAR,
queries=[
BuilderQuery(
signal="traces",
name="A",
filter_expression=f"{field} EXISTS",
aggregations=[Aggregation("count()")],
group_by=[TelemetryFieldKey(requested, "string", context)],
order=[OrderBy(TelemetryFieldKey(requested, "string", context), "asc")],
).to_dict()
],
)
result = _result(grouped)
assert result["columns"][0]["name"] == requested, "response identity must match the request spelling"
assert result["data"] == [["production", 4], ["staging", 1]]
assert _metadata_values(signoz, token, CURRENT, context) == {"production", "staging"}
assert _metadata_values(signoz, token, OLD, context) == {"production", "staging"}
keys_response = requests.get(
signoz.self.host_configs["8080"].get("/api/v1/fields/keys"),
timeout=5,
headers={"authorization": f"Bearer {token}"},
params={"signal": "traces", "searchText": OLD},
)
assert keys_response.status_code == HTTPStatus.OK, keys_response.text
keys = keys_response.json()["data"]["keys"]
assert CURRENT in keys
assert OLD not in keys
start_ns = str(int((now - timedelta(minutes=2)).timestamp() * 1_000_000_000))
end_ns = str(int((now + timedelta(minutes=1)).timestamp() * 1_000_000_000))
for requested in (CURRENT, OLD):
services_response = requests.post(
signoz.self.host_configs["8080"].get("/api/v2/services"),
timeout=30,
headers={"authorization": f"Bearer {token}"},
json={
"start": start_ns,
"end": end_ns,
"tags": [
{
"Key": requested,
"Operator": "In",
"StringValues": ["production"],
"TagType": "ResourceAttribute",
}
],
},
)
assert services_response.status_code == HTTPStatus.OK, services_response.text
services = {item["serviceName"] for item in services_response.json()["data"]}
assert services == PRODUCTION_SPANS
map_response = requests.post(
signoz.self.host_configs["8080"].get("/api/v1/dependency_graph"),
timeout=30,
headers={"authorization": f"Bearer {token}"},
json={
"start": start_ns,
"end": end_ns,
"tags": [
{
"key": requested,
"operator": "In",
"stringValues": ["production"],
"tagType": "ResourceAttribute",
}
],
},
)
assert map_response.status_code == HTTPStatus.OK, map_response.text
assert {edge["parent"] for edge in map_response.json()} == {f"{PREFIX}-map-production"}