Add container-based config discovery support to weaviate

Adds a `from_ports`-based discovery strategy scoped to weaviate's OpenMetrics
port (2112) and wires the kind-based E2E harness to exercise it against a
real Kubernetes cluster via the new kube_discovery helpers.

Environment: Datadog workspace

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

Author: vitkyrka

weaviate/assets/configuration/spec.yaml

@@ -2,6 +2,12 @@ name: Weaviate
 fleet_configurable: true
 files:
 - name: weaviate.yaml
+  discovery:
+    strategies:
+    - template: discovery/openmetrics_from_ports
+      overrides:
+        port_hints:
+        - 2112
   options:
   - template: init_config
     options:
@@ -25,4 +31,12 @@ files:
       value:
         display_default: null
         example: http://localhost:8080
-        type: string
+        type: string
+
+- name: auto_conf.yaml
+  options:
+  - template: ad_identifiers
+    overrides:
+      value.example:
+      - weaviate
+  - template: auto_conf/discovery

weaviate/datadog_checks/weaviate/config_models/discovery.py

@@ -0,0 +1,42 @@
+# (C) Datadog, Inc. 2026-present
+# All rights reserved
+# Licensed under a 3-clause BSD style license (see LICENSE)
+
+# This file is autogenerated.
+# To change this file you should edit assets/configuration/spec.yaml and then run the following commands:
+#     ddev -x validate config -s <INTEGRATION_NAME>
+#     ddev -x validate models -s <INTEGRATION_NAME>
+
+from __future__ import annotations
+
+from collections.abc import Iterator
+from typing import Any
+
+from datadog_checks.base.utils.discovery import Service, candidate_ports
+from datadog_checks.weaviate.config_models import discovery_overrides
+from datadog_checks.weaviate.config_models.instance import InstanceConfig
+from datadog_checks.weaviate.config_models.shared import SharedConfig
+
+
+def _generated_candidates(service: Service) -> Iterator[dict[str, Any]]:
+    shared = SharedConfig.model_validate({}, context={'configured_fields': frozenset()}).model_dump(
+        by_alias=True, mode='json', exclude_none=True
+    )
+    # discovery[0]: from_ports
+    for port in candidate_ports(service, [2112]):
+        ctx = {'port': port}
+        instance_data = {
+            'openmetrics_endpoint': 'http://{service.host}:{port.number}/metrics'.format(service=service, **ctx),
+        }
+        instance = InstanceConfig.model_validate(
+            instance_data, context={'configured_fields': frozenset(instance_data)}
+        ).model_dump(by_alias=True, mode='json', exclude_none=True)
+        yield {'init_config': shared, 'instances': [instance]}
+
+
+def candidates(service: Service) -> Iterator[dict[str, Any]]:
+    override = getattr(discovery_overrides, 'candidates', None)
+    if override is None:
+        yield from _generated_candidates(service)
+    else:
+        yield from override(service, default=_generated_candidates)

weaviate/datadog_checks/weaviate/config_models/discovery_overrides.py

@@ -0,0 +1,12 @@
+# (C) Datadog, Inc. 2026-present
+# All rights reserved
+# Licensed under a 3-clause BSD style license (see LICENSE)
+
+# Override the generated discovery candidates() for this integration.
+#
+# Define a candidates(service, default) function to wrap or replace the generated
+# candidate generation. `default` is the generated generator; call it to reuse
+# the spec-driven candidates, or ignore it to replace them entirely.
+#
+# def candidates(service, default):
+#     yield from default(service)

weaviate/datadog_checks/weaviate/config_models/discovery_strategies.py

@@ -0,0 +1,18 @@
+# (C) Datadog, Inc. 2026-present
+# All rights reserved
+# Licensed under a 3-clause BSD style license (see LICENSE)
+
+# Here you can define custom (local:) discovery strategies for this integration.
+#
+# Decorate a generator with @discovery_strategy (imported from
+# datadog_checks.base.utils.discovery) and reference it from the spec discovery
+# stanza as `strategy: local:<function_name>`. The function receives the
+# discovered Service plus the inputs declared in the spec and yields one context
+# (ctx) mapping per candidate, exposing the keys listed in `provides`.
+#
+# from datadog_checks.base.utils.discovery import discovery_strategy
+#
+# @discovery_strategy(provides=('svc',))
+# def from_some_config(service, config_path):
+#     ...
+#     yield {'svc': ...}

weaviate/datadog_checks/weaviate/data/auto_conf.yaml

@@ -0,0 +1,19 @@
+## @param ad_identifiers - list of strings - required
+## A list of container identifiers that are used by Autodiscovery to identify
+## which container the check should be run against. For more information, see:
+## https://docs.datadoghq.com/agent/guide/ad_identifiers/
+#
+ad_identifiers:
+  - weaviate
+
+## Enables configuration discovery
+#
+discovery: {}
+
+## Unused init configuration
+#
+init_config:
+
+## Unused instance configuration
+#
+instances: []

weaviate/tests/conftest.py

@@ -11,6 +11,7 @@
 
 from datadog_checks.dev import get_here
 from datadog_checks.dev.kind import kind_run
+from datadog_checks.dev.kube_discovery import save_kube_discovery_state, setup_discovery_agent
 from datadog_checks.dev.kube_port_forward import port_forward
 from datadog_checks.dev.subprocess import run_command
 from datadog_checks.weaviate.check import DEFAULT_LIVENESS_ENDPOINT
@@ -37,6 +38,9 @@ def setup_weaviate():
 @pytest.fixture(scope='session')
 def dd_environment():
     with kind_run(conditions=[setup_weaviate]) as kubeconfig, ExitStack() as stack:
+        setup_discovery_agent(kubeconfig)
+        save_kube_discovery_state(kubeconfig)
+
         weaviate_host, weaviate_port = stack.enter_context(
             port_forward(kubeconfig, 'weaviate', 2112, 'statefulset', 'weaviate')
         )

weaviate/tests/test_e2e.py

@@ -4,9 +4,14 @@
 import pytest
 
 from datadog_checks.base.constants import ServiceCheck
+from datadog_checks.dev.kube_discovery import (
+    assert_all_discovery_candidates_stable_kubernetes,
+    run_discovery_check_kubernetes,
+)
 from datadog_checks.dev.utils import get_metadata_metrics
+from datadog_checks.weaviate import WeaviateCheck
 
-from .common import E2E_METRICS, FLAKY_E2E_METRICS
+from .common import E2E_METRICS, FLAKY_E2E_METRICS, OM_METRICS
 
 
 @pytest.mark.e2e
@@ -22,3 +27,31 @@ def test_e2e_openmetrics_v2(dd_agent_check):
 
     aggregator.assert_all_metrics_covered()
     aggregator.assert_metrics_using_metadata(get_metadata_metrics())
+
+
+@pytest.mark.e2e
+def test_e2e_discovery(aggregator, datadog_agent):
+    # Discovery only knows about the metrics port (2112), not the Restful API port (8080), so it
+    # can't populate `weaviate_api_endpoint` — only the OpenMetrics metrics are collected here.
+    run_discovery_check_kubernetes(aggregator, datadog_agent, check_rate=True)
+
+    aggregator.assert_service_check('weaviate.openmetrics.health', ServiceCheck.OK, count=2)
+    for metric in OM_METRICS:
+        if metric in FLAKY_E2E_METRICS:
+            aggregator.assert_metric(metric, at_least=0)
+        else:
+            aggregator.assert_metric(metric)
+
+    aggregator.assert_all_metrics_covered()
+    aggregator.assert_metrics_using_metadata(get_metadata_metrics())
+
+
+@pytest.mark.e2e
+def test_e2e_discovery_all_candidates(aggregator, datadog_agent):
+    assert_all_discovery_candidates_stable_kubernetes(
+        WeaviateCheck,
+        aggregator,
+        datadog_agent,
+        namespace='weaviate',
+        pod_selector='app=weaviate',
+    )