SERVER-122557: Ensure streams js tests are added to a suite (#50997)

GitOrigin-RevId: bd31b4fa0500dae49706cfa6c6d6400475dd15e5

bazel/wrapper_hook/lint.py

@@ -602,6 +602,12 @@ def run_rules_lint(bazel_bin: str, args: list[str]):
         lr.run_bazel("//buildscripts:yamllinters")
         print("No errors found in evergreen yaml")
 
+    if lint_all or any(
+        "jstests/streams" in file or "resmokeconfig/suites/streams" in file
+        for file in files_to_lint
+    ):
+        lr.run_bazel("//buildscripts:streams_suite_coverage_linter")
+
     if lint_all or any(file.endswith(".md") for file in files_to_lint):
         lr.run_bazel("//buildscripts:markdown_link_linter", ["--root=src/mongo", "--verbose"])
 

buildscripts/BUILD.bazel

@@ -522,6 +522,19 @@ py_binary(
     deps = [],  # 'requests' optional (external link checks skipped if absent)
 )
 
+py_binary(
+    name = "streams_suite_coverage_linter",
+    srcs = ["streams_suite_coverage_linter.py"],
+    main = "streams_suite_coverage_linter.py",
+    visibility = ["//visibility:public"],
+    deps = [
+        dependency(
+            "pyyaml",
+            group = "core",
+        ),
+    ],
+)
+
 py_binary(
     name = "yamllinters",
     srcs = ["yamllinters.py"],

buildscripts/resmokeconfig/suites/streams_aspio_iceberg_4.yml

@@ -0,0 +1,21 @@
+test_kind: js_test
+
+selector:
+  roots:
+    - src/mongo/db/modules/*/jstests/streams/aspio/iceberg/iceberg_duplicate_rows.js
+    - src/mongo/db/modules/*/jstests/streams/aspio/iceberg/iceberg_snapshot_lineage.js
+    - src/mongo/db/modules/*/jstests/streams/aspio/iceberg/iceberg_stats.js
+  exclude_files:
+    # TODO(SERVER-117494): Re-enable once flakiness is resolved.
+    - src/mongo/db/modules/*/jstests/streams/aspio/iceberg/gwproxy/iceberg_gwproxy.js
+
+executor:
+  fixture:
+    class: ReplicaSetFixture
+    mongod_options:
+      bind_ip_all: ""
+      set_parameters:
+        enableTestCommands: 1
+        featureFlagStreams: true
+        diagnosticDataCollectionEnabled: false
+    num_nodes: 1

buildscripts/resmokeconfig/suites/streams_aspio_iceberg_large.yml

@@ -8,6 +8,8 @@ selector:
     - src/mongo/db/modules/enterprise/jstests/streams/aspio/iceberg/iceberg_benchmark_tpcc_official.js
     # TODO(SERVER-122755): Move this elsewhere
     - src/mongo/db/modules/enterprise/jstests/streams/aspio/iceberg/s3_emit_benchmark_tpcc.js
+    - src/mongo/db/modules/enterprise/jstests/streams/aspio/iceberg/iceberg_speed_test.js
+    - src/mongo/db/modules/enterprise/jstests/streams/aspio/pubsub/pubsub_benchmark_tpcc.js
 
 executor:
   fixture:

buildscripts/streams_suite_coverage_linter.py

@@ -0,0 +1,141 @@
+#!/usr/bin/env python3
+"""Verify every jstest under streams directories is referenced by at least one resmoke suite.
+
+Exits 0 when all test files are covered, 1 otherwise.
+
+Usage:
+    python3 buildscripts/resmoke_suite_coverage_linter.py
+"""
+
+import fnmatch
+import glob
+import os
+import pathlib
+import sys
+
+import yaml
+
+REPO_ROOT = pathlib.Path(__file__).resolve().parent.parent
+
+ENTERPRISE_MODULE = REPO_ROOT / "src" / "mongo" / "db" / "modules" / "enterprise"
+
+JSTESTS_DIRS = [
+    ENTERPRISE_MODULE / "jstests" / "streams",
+    ENTERPRISE_MODULE / "jstests" / "streams_kafka",
+]
+
+SUITE_DIR = REPO_ROOT / "buildscripts" / "resmokeconfig" / "suites"
+
+SUITE_GLOB = "streams*.yml"
+
+# Filename patterns for non-test helper files that are loaded by tests
+# rather than executed directly by resmoke.
+NON_TEST_PATTERNS: list[str] = [
+    "*_utils.js",
+    "benchmark_utils.js",
+    "*_harness.js",
+    "*_common.js",
+]
+
+# Directories whose contents are never standalone tests (libraries,
+# fixture data, helper functions, etc.).
+NON_TEST_DIRS: set[str] = {
+    "lib",
+    "data",
+    "function",
+    "mongostream_container_manager",
+}
+
+
+def _is_non_test_file(filename: str) -> bool:
+    return any(fnmatch.fnmatch(filename, pat) for pat in NON_TEST_PATTERNS)
+
+
+def _is_in_non_test_dir(path: pathlib.Path, jstests_dir: pathlib.Path) -> bool:
+    rel = path.relative_to(jstests_dir)
+    return any(part in NON_TEST_DIRS for part in rel.parts)
+
+
+def find_all_js_test_files() -> set[str]:
+    """Find all .js files under streams jstests directories, excluding known non-test files.
+
+    Returns paths relative to REPO_ROOT.
+    """
+    files: set[str] = set()
+    for jstests_dir in JSTESTS_DIRS:
+        if not jstests_dir.is_dir():
+            continue
+        for root, _dirs, filenames in os.walk(jstests_dir):
+            root_path = pathlib.Path(root)
+            if _is_in_non_test_dir(root_path, jstests_dir):
+                continue
+            for f in filenames:
+                if not f.endswith(".js"):
+                    continue
+                if _is_non_test_file(f):
+                    continue
+                full = root_path / f
+                files.add(str(full.relative_to(REPO_ROOT)))
+    return files
+
+
+def _normalize_suite_path(path: str) -> str:
+    """Replace the module wildcard with the concrete enterprise path."""
+    return path.replace("modules/*/", "modules/enterprise/")
+
+
+def get_suite_covered_files() -> set[str]:
+    """Parse all streams suite YAMLs and return the set of covered file paths.
+
+    Returns paths relative to REPO_ROOT.
+    """
+    covered: set[str] = set()
+
+    suite_files = sorted(SUITE_DIR.glob(SUITE_GLOB))
+    for suite_file in suite_files:
+        with open(suite_file, encoding="utf-8") as fh:
+            config = yaml.safe_load(fh)
+
+        if not config or "selector" not in config:
+            continue
+
+        selector = config["selector"]
+
+        for root_pattern in selector.get("roots", []):
+            full_pattern = str(REPO_ROOT / _normalize_suite_path(root_pattern))
+            for match in glob.glob(full_pattern):
+                covered.add(os.path.relpath(match, REPO_ROOT))
+
+        for excl_pattern in selector.get("exclude_files", []):
+            full_pattern = str(REPO_ROOT / _normalize_suite_path(excl_pattern))
+            for match in glob.glob(full_pattern):
+                covered.add(os.path.relpath(match, REPO_ROOT))
+
+    return covered
+
+
+def main() -> int:
+    all_tests = find_all_js_test_files()
+    covered = get_suite_covered_files()
+    uncovered = sorted(all_tests - covered)
+
+    if not uncovered:
+        print("All streams jstests are covered by resmoke suites!")
+        return 0
+
+    print(f"Found {len(uncovered)} streams jstest(s) not referenced by any resmoke suite:\n")
+    for f in uncovered:
+        print(f"  {f}")
+    print(
+        "\nPlease add each test to the appropriate streams*.yml suite config in\n"
+        "buildscripts/resmokeconfig/suites/\n"
+        "\n"
+        "If a file is not a standalone test (e.g. a utility loaded by other tests),\n"
+        "add its filename pattern to NON_TEST_PATTERNS or NON_TEST_DIRS in\n"
+        "buildscripts/streams_suite_coverage_linter.py"
+    )
+    return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())

etc/burn_in_tests.yml

@@ -14,6 +14,7 @@ selector:
       - streams_aspio_iceberg_1
       - streams_aspio_iceberg_2
       - streams_aspio_iceberg_3
+      - streams_aspio_iceberg_4
       - streams_aspio_iceberg_large
       - streams_aspio_iceberg_5
       - streams_aspio_utilities

etc/evergreen_lint.yml

@@ -185,6 +185,7 @@ rules:
             "streams_aspio_iceberg_1",
             "streams_aspio_iceberg_2",
             "streams_aspio_iceberg_3",
+            "streams_aspio_iceberg_4",
             "streams_aspio_iceberg_large",
             "streams_aspio_iceberg_5",
             "streams_aspio_utilities",

etc/evergreen_yml_components/tasks/resmoke/non_server_teams/tasks.yml

@@ -491,6 +491,21 @@ tasks:
         vars:
           resmoke_jobs_max: 4
 
+  - <<: *task_template_streams
+    name: streams_aspio_iceberg_4
+    tags:
+      [
+        "assigned_to_jira_team_streams",
+        "experimental",
+        "streams_release_test",
+        "requires_extra_system_deps",
+      ]
+    commands:
+      - func: "do setup for streams"
+      - func: "run streams tests"
+        vars:
+          resmoke_jobs_max: 4
+
   - <<: *task_template_streams
     name: streams_aspio_iceberg_large
     tags: