summaryrefslogtreecommitdiffstats
path: root/taskcluster/gecko_taskgraph/test/test_optimize_strategies.py
diff options
context:
space:
mode:
Diffstat (limited to 'taskcluster/gecko_taskgraph/test/test_optimize_strategies.py')
-rw-r--r--taskcluster/gecko_taskgraph/test/test_optimize_strategies.py551
1 files changed, 551 insertions, 0 deletions
diff --git a/taskcluster/gecko_taskgraph/test/test_optimize_strategies.py b/taskcluster/gecko_taskgraph/test/test_optimize_strategies.py
new file mode 100644
index 0000000000..0f37332300
--- /dev/null
+++ b/taskcluster/gecko_taskgraph/test/test_optimize_strategies.py
@@ -0,0 +1,551 @@
+# Any copyright is dedicated to the public domain.
+# http://creativecommons.org/publicdomain/zero/1.0/
+
+
+import time
+from datetime import datetime
+from time import mktime
+
+import pytest
+from mozunit import main
+from taskgraph.optimize.base import registry
+from taskgraph.task import Task
+
+from gecko_taskgraph.optimize import project
+from gecko_taskgraph.optimize.backstop import SkipUnlessBackstop, SkipUnlessPushInterval
+from gecko_taskgraph.optimize.bugbug import (
+ FALLBACK,
+ BugBugPushSchedules,
+ DisperseGroups,
+ SkipUnlessDebug,
+)
+from gecko_taskgraph.optimize.strategies import IndexSearch, SkipUnlessSchedules
+from gecko_taskgraph.util.backstop import BACKSTOP_PUSH_INTERVAL
+from gecko_taskgraph.util.bugbug import (
+ BUGBUG_BASE_URL,
+ BugbugTimeoutException,
+ push_schedules,
+)
+
+
+@pytest.fixture(autouse=True)
+def clear_push_schedules_memoize():
+ push_schedules.clear()
+
+
+@pytest.fixture
+def params():
+ return {
+ "branch": "autoland",
+ "head_repository": "https://hg.mozilla.org/integration/autoland",
+ "head_rev": "abcdef",
+ "project": "autoland",
+ "pushlog_id": 1,
+ "pushdate": mktime(datetime.now().timetuple()),
+ }
+
+
+def generate_tasks(*tasks):
+ for i, task in enumerate(tasks):
+ task.setdefault("label", f"task-{i}-label")
+ task.setdefault("kind", "test")
+ task.setdefault("task", {})
+ task.setdefault("attributes", {})
+ task["attributes"].setdefault("e10s", True)
+
+ for attr in (
+ "optimization",
+ "dependencies",
+ "soft_dependencies",
+ ):
+ task.setdefault(attr, None)
+
+ task["task"].setdefault("label", task["label"])
+ yield Task.from_json(task)
+
+
+# task sets
+
+default_tasks = list(
+ generate_tasks(
+ {"attributes": {"test_manifests": ["foo/test.ini", "bar/test.ini"]}},
+ {"attributes": {"test_manifests": ["bar/test.ini"], "build_type": "debug"}},
+ {"attributes": {"build_type": "debug"}},
+ {"attributes": {"test_manifests": [], "build_type": "opt"}},
+ {"attributes": {"build_type": "opt"}},
+ )
+)
+
+
+disperse_tasks = list(
+ generate_tasks(
+ {
+ "attributes": {
+ "test_manifests": ["foo/test.ini", "bar/test.ini"],
+ "test_platform": "linux/opt",
+ }
+ },
+ {
+ "attributes": {
+ "test_manifests": ["bar/test.ini"],
+ "test_platform": "linux/opt",
+ }
+ },
+ {
+ "attributes": {
+ "test_manifests": ["bar/test.ini"],
+ "test_platform": "windows/debug",
+ }
+ },
+ {
+ "attributes": {
+ "test_manifests": ["bar/test.ini"],
+ "test_platform": "linux/opt",
+ "unittest_variant": "no-fission",
+ }
+ },
+ {
+ "attributes": {
+ "e10s": False,
+ "test_manifests": ["bar/test.ini"],
+ "test_platform": "linux/opt",
+ }
+ },
+ )
+)
+
+
+def idfn(param):
+ if isinstance(param, tuple):
+ try:
+ return param[0].__name__
+ except AttributeError:
+ return None
+ return None
+
+
+@pytest.mark.parametrize(
+ "opt,tasks,arg,expected",
+ [
+ # debug
+ pytest.param(
+ SkipUnlessDebug(),
+ default_tasks,
+ None,
+ ["task-0-label", "task-1-label", "task-2-label"],
+ ),
+ # disperse with no supplied importance
+ pytest.param(
+ DisperseGroups(),
+ disperse_tasks,
+ None,
+ [t.label for t in disperse_tasks],
+ ),
+ # disperse with low importance
+ pytest.param(
+ DisperseGroups(),
+ disperse_tasks,
+ {"bar/test.ini": "low"},
+ ["task-0-label", "task-2-label"],
+ ),
+ # disperse with medium importance
+ pytest.param(
+ DisperseGroups(),
+ disperse_tasks,
+ {"bar/test.ini": "medium"},
+ ["task-0-label", "task-1-label", "task-2-label"],
+ ),
+ # disperse with high importance
+ pytest.param(
+ DisperseGroups(),
+ disperse_tasks,
+ {"bar/test.ini": "high"},
+ ["task-0-label", "task-1-label", "task-2-label", "task-3-label"],
+ ),
+ ],
+ ids=idfn,
+)
+def test_optimization_strategy_remove(params, opt, tasks, arg, expected):
+ labels = [t.label for t in tasks if not opt.should_remove_task(t, params, arg)]
+ assert sorted(labels) == sorted(expected)
+
+
+@pytest.mark.parametrize(
+ "state,expires,expected",
+ (
+ ("completed", "2021-06-06T14:53:16.937Z", False),
+ ("completed", "2021-06-08T14:53:16.937Z", "abc"),
+ ("exception", "2021-06-08T14:53:16.937Z", False),
+ ("failed", "2021-06-08T14:53:16.937Z", False),
+ ),
+)
+def test_index_search(responses, params, state, expires, expected):
+ taskid = "abc"
+ index_path = "foo.bar.latest"
+ responses.add(
+ responses.GET,
+ f"https://firefox-ci-tc.services.mozilla.com/api/index/v1/task/{index_path}",
+ json={"taskId": taskid},
+ status=200,
+ )
+
+ responses.add(
+ responses.GET,
+ f"https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/{taskid}/status",
+ json={
+ "status": {
+ "state": state,
+ "expires": expires,
+ }
+ },
+ status=200,
+ )
+
+ opt = IndexSearch()
+ deadline = "2021-06-07T19:03:20.482Z"
+ assert opt.should_replace_task({}, params, deadline, (index_path,)) == expected
+
+
+@pytest.mark.parametrize(
+ "args,data,expected",
+ [
+ # empty
+ pytest.param(
+ (0.1,),
+ {},
+ [],
+ ),
+ # only tasks without test manifests selected
+ pytest.param(
+ (0.1,),
+ {"tasks": {"task-1-label": 0.9, "task-2-label": 0.1, "task-3-label": 0.5}},
+ ["task-2-label"],
+ ),
+ # tasks which are unknown to bugbug are selected
+ pytest.param(
+ (0.1,),
+ {
+ "tasks": {"task-1-label": 0.9, "task-3-label": 0.5},
+ "known_tasks": ["task-1-label", "task-3-label", "task-4-label"],
+ },
+ ["task-2-label"],
+ ),
+ # tasks containing groups selected
+ pytest.param(
+ (0.1,),
+ {"groups": {"foo/test.ini": 0.4}},
+ ["task-0-label"],
+ ),
+ # tasks matching "tasks" or "groups" selected
+ pytest.param(
+ (0.1,),
+ {
+ "tasks": {"task-2-label": 0.2},
+ "groups": {"foo/test.ini": 0.25, "bar/test.ini": 0.75},
+ },
+ ["task-0-label", "task-1-label", "task-2-label"],
+ ),
+ # tasks matching "tasks" or "groups" selected, when they exceed the confidence threshold
+ pytest.param(
+ (0.5,),
+ {
+ "tasks": {"task-2-label": 0.2, "task-4-label": 0.5},
+ "groups": {"foo/test.ini": 0.65, "bar/test.ini": 0.25},
+ },
+ ["task-0-label", "task-4-label"],
+ ),
+ # tasks matching "reduced_tasks" are selected, when they exceed the confidence threshold
+ pytest.param(
+ (0.7, True, True),
+ {
+ "tasks": {"task-2-label": 0.7, "task-4-label": 0.7},
+ "reduced_tasks": {"task-4-label": 0.7},
+ "groups": {"foo/test.ini": 0.75, "bar/test.ini": 0.25},
+ },
+ ["task-4-label"],
+ ),
+ # tasks matching "groups" selected, only on specific platforms.
+ pytest.param(
+ (0.1, False, False, None, 1, True),
+ {
+ "tasks": {"task-2-label": 0.2},
+ "groups": {"foo/test.ini": 0.25, "bar/test.ini": 0.75},
+ "config_groups": {
+ "foo/test.ini": ["task-1-label", "task-0-label"],
+ "bar/test.ini": ["task-0-label"],
+ },
+ },
+ ["task-0-label", "task-2-label"],
+ ),
+ pytest.param(
+ (0.1, False, False, None, 1, True),
+ {
+ "tasks": {"task-2-label": 0.2},
+ "groups": {"foo/test.ini": 0.25, "bar/test.ini": 0.75},
+ "config_groups": {
+ "foo/test.ini": ["task-1-label", "task-0-label"],
+ "bar/test.ini": ["task-1-label"],
+ },
+ },
+ ["task-0-label", "task-1-label", "task-2-label"],
+ ),
+ pytest.param(
+ (0.1, False, False, None, 1, True),
+ {
+ "tasks": {"task-2-label": 0.2},
+ "groups": {"foo/test.ini": 0.25, "bar/test.ini": 0.75},
+ "config_groups": {
+ "foo/test.ini": ["task-1-label"],
+ "bar/test.ini": ["task-0-label"],
+ },
+ },
+ ["task-0-label", "task-2-label"],
+ ),
+ pytest.param(
+ (0.1, False, False, None, 1, True),
+ {
+ "tasks": {"task-2-label": 0.2},
+ "groups": {"foo/test.ini": 0.25, "bar/test.ini": 0.75},
+ "config_groups": {
+ "foo/test.ini": ["task-1-label"],
+ "bar/test.ini": ["task-3-label"],
+ },
+ },
+ ["task-2-label"],
+ ),
+ ],
+ ids=idfn,
+)
+def test_bugbug_push_schedules(responses, params, args, data, expected):
+ query = "/push/{branch}/{head_rev}/schedules".format(**params)
+ url = BUGBUG_BASE_URL + query
+
+ responses.add(
+ responses.GET,
+ url,
+ json=data,
+ status=200,
+ )
+
+ opt = BugBugPushSchedules(*args)
+ labels = [
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ ]
+ assert sorted(labels) == sorted(expected)
+
+
+def test_bugbug_multiple_pushes(responses, params):
+ pushes = {str(pid): {"changesets": [f"c{pid}"]} for pid in range(8, 10)}
+
+ responses.add(
+ responses.GET,
+ "https://hg.mozilla.org/integration/autoland/json-pushes/?version=2&startID=8&endID=9",
+ json={"pushes": pushes},
+ status=200,
+ )
+
+ responses.add(
+ responses.GET,
+ BUGBUG_BASE_URL + "/push/{}/c9/schedules".format(params["branch"]),
+ json={
+ "tasks": {"task-2-label": 0.2, "task-4-label": 0.5},
+ "groups": {"foo/test.ini": 0.2, "bar/test.ini": 0.25},
+ "config_groups": {"foo/test.ini": ["linux-*"], "bar/test.ini": ["task-*"]},
+ "known_tasks": ["task-4-label"],
+ },
+ status=200,
+ )
+
+ # Tasks with a lower confidence don't override task with a higher one.
+ # Tasks with a higher confidence override tasks with a lower one.
+ # Known tasks are merged.
+ responses.add(
+ responses.GET,
+ BUGBUG_BASE_URL + "/push/{branch}/{head_rev}/schedules".format(**params),
+ json={
+ "tasks": {"task-2-label": 0.2, "task-4-label": 0.2},
+ "groups": {"foo/test.ini": 0.65, "bar/test.ini": 0.25},
+ "config_groups": {
+ "foo/test.ini": ["task-*"],
+ "bar/test.ini": ["windows-*"],
+ },
+ "known_tasks": ["task-1-label", "task-3-label"],
+ },
+ status=200,
+ )
+
+ params["pushlog_id"] = 10
+
+ opt = BugBugPushSchedules(0.3, False, False, False, 2)
+ labels = [
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ ]
+ assert sorted(labels) == sorted(["task-0-label", "task-2-label", "task-4-label"])
+
+ opt = BugBugPushSchedules(0.3, False, False, False, 2, True)
+ labels = [
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ ]
+ assert sorted(labels) == sorted(["task-0-label", "task-2-label", "task-4-label"])
+
+ opt = BugBugPushSchedules(0.2, False, False, False, 2, True)
+ labels = [
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ ]
+ assert sorted(labels) == sorted(
+ ["task-0-label", "task-1-label", "task-2-label", "task-4-label"]
+ )
+
+
+def test_bugbug_timeout(monkeypatch, responses, params):
+ query = "/push/{branch}/{head_rev}/schedules".format(**params)
+ url = BUGBUG_BASE_URL + query
+ responses.add(
+ responses.GET,
+ url,
+ json={"ready": False},
+ status=202,
+ )
+
+ # Make sure the test runs fast.
+ monkeypatch.setattr(time, "sleep", lambda i: None)
+
+ opt = BugBugPushSchedules(0.5)
+ with pytest.raises(BugbugTimeoutException):
+ opt.should_remove_task(default_tasks[0], params, None)
+
+
+def test_bugbug_fallback(monkeypatch, responses, params):
+ query = "/push/{branch}/{head_rev}/schedules".format(**params)
+ url = BUGBUG_BASE_URL + query
+ responses.add(
+ responses.GET,
+ url,
+ json={"ready": False},
+ status=202,
+ )
+
+ opt = BugBugPushSchedules(0.5, fallback=FALLBACK)
+
+ # Make sure the test runs fast.
+ monkeypatch.setattr(time, "sleep", lambda i: None)
+
+ def fake_should_remove_task(task, params, _):
+ return task.label == default_tasks[0].label
+
+ monkeypatch.setattr(
+ registry[FALLBACK], "should_remove_task", fake_should_remove_task
+ )
+
+ assert opt.should_remove_task(default_tasks[0], params, None)
+
+ # Make sure we don't hit bugbug more than once.
+ responses.reset()
+
+ assert not opt.should_remove_task(default_tasks[1], params, None)
+
+
+def test_backstop(params):
+ all_labels = {t.label for t in default_tasks}
+ opt = SkipUnlessBackstop()
+
+ params["backstop"] = False
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == set()
+
+ params["backstop"] = True
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == all_labels
+
+
+def test_push_interval(params):
+ all_labels = {t.label for t in default_tasks}
+ opt = SkipUnlessPushInterval(10) # every 10th push
+
+ # Only multiples of 10 schedule tasks.
+ params["pushlog_id"] = 9
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == set()
+
+ params["pushlog_id"] = 10
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == all_labels
+
+
+def test_expanded(params):
+ all_labels = {t.label for t in default_tasks}
+ opt = registry["skip-unless-expanded"]
+
+ params["backstop"] = False
+ params["pushlog_id"] = BACKSTOP_PUSH_INTERVAL / 2
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == all_labels
+
+ params["pushlog_id"] += 1
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == set()
+
+ params["backstop"] = True
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, None)
+ }
+ assert scheduled == all_labels
+
+
+def test_project_autoland_test(monkeypatch, responses, params):
+ """Tests the behaviour of the `project.autoland["test"]` strategy on
+ various types of pushes.
+ """
+ # This is meant to test the composition of substrategies, and not the
+ # actual optimization implementations. So mock them out for simplicity.
+ monkeypatch.setattr(SkipUnlessSchedules, "should_remove_task", lambda *args: False)
+ monkeypatch.setattr(DisperseGroups, "should_remove_task", lambda *args: False)
+
+ def fake_bugbug_should_remove_task(self, task, params, importance):
+ if self.num_pushes > 1:
+ return task.label == "task-4-label"
+ return task.label in ("task-2-label", "task-3-label", "task-4-label")
+
+ monkeypatch.setattr(
+ BugBugPushSchedules, "should_remove_task", fake_bugbug_should_remove_task
+ )
+
+ opt = project.autoland["test"]
+
+ # On backstop pushes, nothing gets optimized.
+ params["backstop"] = True
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ }
+ assert scheduled == {t.label for t in default_tasks}
+
+ # On expanded pushes, some things are optimized.
+ params["backstop"] = False
+ params["pushlog_id"] = 10
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ }
+ assert scheduled == {"task-0-label", "task-1-label", "task-2-label", "task-3-label"}
+
+ # On regular pushes, more things are optimized.
+ params["pushlog_id"] = 11
+ scheduled = {
+ t.label for t in default_tasks if not opt.should_remove_task(t, params, {})
+ }
+ assert scheduled == {"task-0-label", "task-1-label"}
+
+
+if __name__ == "__main__":
+ main()