Merge branch 'main' into may_01_rle

VERSION

@@ -1 +1 @@
-24.4.1
+24.5.1

celery_config.py

@@ -106,9 +106,10 @@ def init_celery_tracing(*args, **kwargs):
 hourly_check_task_name = "app.cron.hourly_check.HourlyCheckTask"
 daily_plan_manager_task_name = "app.cron.daily.PlanManagerTask"
 
-backfill_gh_app_installations_name = (
-    "app.tasks.backfill_gh_app_installations.BackfillGHAppInstallationsTask"
-)
+# Backfill GH Apps
+backfill_existing_gh_app_installations_name = "app.tasks.backfill_existing_gh_app_installations.BackfillExistingGHAppInstallationsTask"
+backfill_owners_without_gh_app_installations_name = "app.tasks.backfill_owners_without_gh_app_installations.BackfillOwnersWithoutGHAppInstallationsTask"
+
 trial_expiration_task_name = "app.tasks.plan.TrialExpirationTask"
 trial_expiration_cron_task_name = "app.cron.plan.TrialExpirationCronTask"
 

conftest.py

@@ -325,7 +325,7 @@ def mock_feature(mocker, request):
 
     from shared.rollouts import Feature
 
-    def check_value(self, *, owner_id=None, repo_id=None, default=False):
+    def check_value(self, identifier, default=False):
         return default
 
     return mocker.patch.object(Feature, "check_value", check_value)

database/enums.py

@@ -68,3 +68,7 @@ class FlakeSymptomType(Enum):
     FAILED_IN_DEFAULT_BRANCH = "failed_in_default_branch"
     CONSECUTIVE_DIFF_OUTCOMES = "consecutive_diff_outcomes"
     UNRELATED_MATCHING_FAILURES = "unrelated_matching_failures"
+
+
+class TestResultsProcessingError(Enum):
+    NO_SUCCESS = "no_success"

database/models/reports.py

@@ -312,3 +312,4 @@ class TestResultReportTotals(CodecovBaseModel, MixinBaseClass):
     passed = Column(types.Integer)
     skipped = Column(types.Integer)
     failed = Column(types.Integer)
+    error = Column(types.String(100), nullable=True)

helpers/backfills.py

@@ -0,0 +1,67 @@
+import logging
+
+import shared.torngit as torngit
+from asgiref.sync import async_to_sync
+from sqlalchemy.orm.session import Session
+
+from database.models.core import GithubAppInstallation, Repository
+
+log = logging.getLogger(__name__)
+
+# GH App Backfills
+# Looping and adding all repositories in the installation app
+def add_repos_service_ids_from_provider(
+    db_session: Session,
+    ownerid: int,
+    owner_service: torngit.base.TorngitBaseAdapter,
+    gh_app_installation: GithubAppInstallation,
+):
+    repos = async_to_sync(owner_service.list_repos_using_installation)()
+
+    if repos:
+        # Fetching all repos service ids we have for that owner in the DB
+        repo_service_ids_in_db = [
+            repo.service_id
+            for repo in db_session.query(Repository.service_id)
+            .filter_by(ownerid=ownerid)
+            .all()
+        ]
+
+        # Add service ids from provider that we have DB records for to a list
+        new_repo_service_ids = []
+        for repo in repos:
+            repo_data = repo["repo"]
+            service_id = repo_data["service_id"]
+            if service_id and service_id in repo_service_ids_in_db:
+                new_repo_service_ids.append(service_id)
+        log.info(
+            "Added the following repo service ids to this gh app installation",
+            extra=dict(
+                ownerid=ownerid,
+                installation_id=gh_app_installation.installation_id,
+                new_repo_service_ids=new_repo_service_ids,
+            ),
+        )
+        gh_app_installation.repository_service_ids = new_repo_service_ids
+        db_session.commit()
+
+
+# Check if gh selection is set to all and act accordingly
+def maybe_set_installation_to_all_repos(
+    db_session: Session,
+    owner_service,
+    gh_app_installation: GithubAppInstallation,
+):
+    remote_gh_app_installation = async_to_sync(owner_service.get_gh_app_installation)(
+        installation_id=gh_app_installation.installation_id
+    )
+    repository_selection = remote_gh_app_installation.get("repository_selection", "")
+    if repository_selection == "all":
+        gh_app_installation.repository_service_ids = None
+        db_session.commit()
+        log.info(
+            "Selection is set to all, no installation is needed",
+            extra=dict(ownerid=gh_app_installation.ownerid),
+        )
+        return True
+    return False

helpers/checkpoint_logger/flows.py

@@ -19,7 +19,11 @@
     "NOTIF_ERROR_NO_REPORT",
 )
 @success_events(
-    "SKIPPING_NOTIFICATION", "NOTIFIED", "NO_PENDING_JOBS", "NOTIF_STALE_HEAD"
+    "SKIPPING_NOTIFICATION",
+    "NOTIFIED",
+    "NO_PENDING_JOBS",
+    "NOTIF_STALE_HEAD",
+    "NO_REPORTS_FOUND",
 )
 @subflows(
     ("time_before_processing", "UPLOAD_TASK_BEGIN", "PROCESSING_BEGIN"),
@@ -36,6 +40,7 @@
 class UploadFlow(BaseFlow):
     UPLOAD_TASK_BEGIN = auto()
     NO_PENDING_JOBS = auto()
+    NO_REPORTS_FOUND = auto()
     TOO_MANY_RETRIES = auto()
     PROCESSING_BEGIN = auto()
     INITIAL_PROCESSING_COMPLETE = auto()

helpers/sentry.py

@@ -41,3 +41,5 @@ def initialize_sentry() -> None:
         ],
         release=os.getenv("SENTRY_RELEASE", version_str),
     )
+    if os.getenv("CLUSTER_ENV"):
+        sentry_sdk.set_tag("cluster", os.getenv("CLUSTER_ENV"))

helpers/telemetry.py

@@ -1,8 +1,10 @@
 import asyncio
+import logging
 from datetime import datetime
 
 import django
 from asgiref.sync import sync_to_async
+from psycopg2 import OperationalError
 from shared.django_apps.pg_telemetry.models import SimpleMetric as PgSimpleMetric
 from shared.django_apps.ts_telemetry.models import SimpleMetric as TsSimpleMetric
 
@@ -11,6 +13,8 @@
 
 from .timeseries import timeseries_enabled
 
+log = logging.getLogger(__name__)
+
 
 def fire_and_forget(fn):
     """
@@ -131,14 +135,19 @@ def log_simple_metric(self, name: str, value: float):
         )
 
         if timeseries_enabled():
-            TsSimpleMetric.objects.create(
-                timestamp=timestamp,
-                name=name,
-                value=value,
-                repo_slug=self.repo_slug,
-                owner_slug=self.owner_slug,
-                commit_slug=self.commit_slug,
-            )
+            try:
+                TsSimpleMetric.objects.create(
+                    timestamp=timestamp,
+                    name=name,
+                    value=value,
+                    repo_slug=self.repo_slug,
+                    owner_slug=self.owner_slug,
+                    commit_slug=self.commit_slug,
+                )
+            except OperationalError:
+                log.warning(
+                    "Failed to create TsSimpleMetric object, Timescale may be unavailable. However we will continue the current task."
+                )
 
     @fire_and_forget
     async def attempt_log_simple_metric(self, name: str, value: float):

helpers/tests/unit/test_sentry.py

@@ -6,8 +6,13 @@
 class TestSentry(object):
     def test_initialize_sentry(self, mocker, mock_configuration):
         mock_configuration._params["services"] = {"sentry": {"server_dsn": "this_dsn"}}
-        mocker.patch.dict(os.environ, {"RELEASE_VERSION": "FAKE_VERSION_FOR_YOU"})
+        cluster = "test_env"
+        mocker.patch.dict(
+            os.environ,
+            {"RELEASE_VERSION": "FAKE_VERSION_FOR_YOU", "CLUSTER_ENV": cluster},
+        )
         mocked_init = mocker.patch("helpers.sentry.sentry_sdk.init")
+        mocked_set_tag = mocker.patch("helpers.sentry.sentry_sdk.set_tag")
         assert initialize_sentry() is None
         mocked_init.assert_called_with(
             "this_dsn",
@@ -18,3 +23,4 @@ def test_initialize_sentry(self, mocker, mock_configuration):
             environment="production",
             integrations=[mocker.ANY, mocker.ANY, mocker.ANY, mocker.ANY],
         )
+        mocked_set_tag.assert_called_with("cluster", cluster)

requirements.in

@@ -1,4 +1,4 @@
-https://github.com/codecov/shared/archive/a0b5d256b6d580a99dd3a1e900f5c9bdaf2d5c9e.tar.gz#egg=shared
+https://github.com/codecov/shared/archive/039efcfca9c11a5d6d02e1e4d19b86d022a21a72.tar.gz#egg=shared
 https://github.com/codecov/opentelem-python/archive/refs/tags/v0.0.4a1.tar.gz#egg=codecovopentelem
 https://github.com/codecov/test-results-parser/archive/5515e960d5d38881036e9127f86320efca649f13.tar.gz#egg=test-results-parser
 boto3>=1.34

requirements.txt

@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.9
+# This file is autogenerated by pip-compile with Python 3.12
 # by the following command:
 #
 #    pip-compile requirements.in
@@ -18,8 +18,6 @@ anyio==3.6.1
     #   openai
 asgiref==3.7.2
     # via django
-async-timeout==4.0.3
-    # via redis
 backoff==1.6.0
     # via analytics-python
 billiard==4.2.0
@@ -119,8 +117,6 @@ ecdsa==0.18.0
     # via tlslite-ng
 excel-base==1.0.4
     # via django-excel-response2
-exceptiongroup==1.2.1
-    # via pytest
 factory-boy==3.2.0
     # via -r requirements.in
 faker==8.8.2
@@ -361,9 +357,7 @@ requests==2.31.0
 respx==0.20.2
     # via -r requirements.in
 rfc3986[idna2008]==1.4.0
-    # via
-    #   httpx
-    #   rfc3986
+    # via httpx
 rsa==4.7.2
     # via google-auth
 s3transfer==0.10.1
@@ -374,7 +368,7 @@ sentry-sdk==1.40.0
     # via
     #   -r requirements.in
     #   shared
-shared @ https://github.com/codecov/shared/archive/a0b5d256b6d580a99dd3a1e900f5c9bdaf2d5c9e.tar.gz
+shared @ https://github.com/codecov/shared/archive/039efcfca9c11a5d6d02e1e4d19b86d022a21a72.tar.gz
     # via -r requirements.in
 six==1.16.0
     # via
@@ -418,18 +412,12 @@ timestring @ https://github.com/codecov/timestring/archive/d37ceacc5954dff3b5bd2
     # via -r requirements.in
 tlslite-ng==0.8.0b1
     # via shared
-tomli==2.0.1
-    # via
-    #   coverage
-    #   pytest
 tqdm==4.66.1
     # via openai
 typing==3.7.4.3
     # via shared
 typing-extensions==4.6.3
     # via
-    #   asgiref
-    #   kombu
     #   openai
     #   opentelemetry-sdk
     #   pydantic

services/bots.py

@@ -38,60 +38,87 @@ def _get_installation_weight(installation: GithubAppInstallation) -> int:
         return MAX_GITHUB_APP_SELECTION_WEIGHT
     seconds_in_hour = 3600
     age_hours = (age.seconds // seconds_in_hour) + age.days * 24
-    return age_hours + 2**age.days
+    # Prevent clock differences from making the weight negative
+    return max(1, age_hours + 2**age.days)
+
+
+def _can_use_this_app(
+    app: GithubAppInstallation, installation_name: str, repository: Optional[Repository]
+) -> bool:
+    return (
+        app.name == installation_name
+        # We ignore apps that are not configured because those can't be used
+        and app.is_configured()
+        and (
+            # If there is a repo we want only the apps that cover said repo
+            (repository and app.is_repo_covered_by_integration(repository))
+            # If there is no repo we still need some true value
+            or (not repository)
+        )
+    )
 
 
 def _get_apps_from_weighted_selection(
     owner: Owner, installation_name: str, repository: Optional[Repository]
 ) -> List[GithubAppInstallation]:
-    ghapp_installations_filter: List[GithubAppInstallation] = list(
-        filter(
-            lambda obj: (
-                obj.name == installation_name
-                and obj.is_configured()
-                and (
-                    # If there is a repo we want only the apps that cover said repo
-                    (repository and obj.is_repo_covered_by_integration(repository))
-                    # If there is no repo we still need some true value
-                    or (not repository)
-                )
-            ),
+    """This function returns an ordered list of GithubAppInstallations that can be used to communicate with GitHub
+    in behalf of the owner. The list is ordered in such a way that the 1st element is the app to be used in Torngit,
+    and the subsequent apps are selected as fallbacks.
+
+    IF the repository is provided, the selected apps also cover the repo.
+    IF installation_name is not the default one, than the default codecov installation
+      is also selected as a possible fallback app.
+
+    Apps are selected randomly but assigned weights based on how recently they were created.
+    This means that older apps are selected more frequently as the main app than newer ones.
+    (up to 10 days, when the probability of being chosen is the same)
+    The random selection is done so we can distribute request load more evenly among apps.
+    """
+    # Map GithubAppInstallation.id --> GithubAppInstallation
+    ghapp_installations_filter: Dict[int, GithubAppInstallation] = {
+        obj.id: obj
+        for obj in filter(
+            lambda obj: _can_use_this_app(obj, installation_name, repository),
             owner.github_app_installations or [],
         )
-    )
+    }
     # We assign weights to the apps based on how long ago they were created.
     # The idea is that there's a greater chance that a change misconfigured the app,
     # So apps recently created are selected less frequently than older apps
+    keys = list(ghapp_installations_filter.keys())
     weights = [
-        min(MAX_GITHUB_APP_SELECTION_WEIGHT, _get_installation_weight(obj))
-        for obj in ghapp_installations_filter
+        min(
+            MAX_GITHUB_APP_SELECTION_WEIGHT,
+            _get_installation_weight(ghapp_installations_filter[key]),
+        )
+        for key in keys
     ]
-    # Random selection of size 3.
-    # If all apps have roughly the same probability of being selected, the array would have different entries.
-    # If 1 app dominates the probability of selection than it would probably be that app repeated 3 times, BUT
-    # from time to time the less frequent one would be selected.
-    apps_to_consider = (
-        random.choices(ghapp_installations_filter, weights=weights, k=3)
-        if len(ghapp_installations_filter) > 0
-        else []
-    )
+    # We pick apps one by one until all apps have been selected
+    # Obviously apps with a higher weight have a higher change of being selected as the main app (1st selection)
+    # But it's important that others are also selected so we can use them as fallbacks
+    apps_to_consider = []
+    apps_to_select = len(keys)
+    selections = 0
+    while selections < apps_to_select:
+        selected_app_id = random.choices(keys, weights, k=1)[0]
+        apps_to_consider.append(ghapp_installations_filter[selected_app_id])
+        # random.choices chooses with replacement
+        # which we are trying to avoid here. So we remove the key selected and its weight from the population.
+        key_idx = keys.index(selected_app_id)
+        keys.pop(key_idx)
+        weights.pop(key_idx)
+        selections += 1
     if installation_name != GITHUB_APP_INSTALLATION_DEFAULT_NAME:
         # Add the default app as the last fallback if the owner is using a different app for the task
         default_apps = filter(
-            lambda obj: obj.name == GITHUB_APP_INSTALLATION_DEFAULT_NAME,
+            lambda obj: _can_use_this_app(
+                obj, GITHUB_APP_INSTALLATION_DEFAULT_NAME, repository
+            ),
             owner.github_app_installations,
         )
         if default_apps:
             apps_to_consider.extend(default_apps)
-    # Now we de-duplicate the apps_to_consider list before returning
-    seen_ids = dict()
-    list_to_return = []
-    for app in apps_to_consider:
-        if seen_ids.get(app.id, False):
-            continue
-        seen_ids[app.id] = True
-        list_to_return.append(app)
-    return list_to_return
+    return apps_to_consider
 
 
 def get_owner_installation_id(