feat(repos): Add task to periodically sync repos from github

This adds a task that runs every 24 hours to make sure that the available repositories on the github side match the repositories we have on the Sentry side.

Author: wedamija

src/sentry/conf/server.py

@@ -1248,6 +1248,10 @@ def SOCIAL_AUTH_DEFAULT_USERNAME() -> str:
         "task": "sdk.control:sentry.tasks.release_registry.fetch_release_registry_data_control",
         "schedule": crontab("*/5", "*", "*", "*", "*"),
     },
+    "github-repo-sync-beat": {
+        "task": "integrations.control:sentry.integrations.github.tasks.sync_repos.github_repo_sync_beat",
+        "schedule": timedelta(minutes=1),
+    },
 }
 
 if SILO_MODE == "CONTROL":

src/sentry/features/temporary.py

@@ -137,6 +137,7 @@ def register_temporary_features(manager: FeatureManager) -> None:
     manager.add("organizations:integrations-github-copilot-agent", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=True)
     manager.add("organizations:integrations-github-platform-detection", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=True)
     manager.add("organizations:github-repo-auto-sync", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=False)
+    manager.add("organizations:github-repo-auto-sync-apply", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=False)
     manager.add("organizations:integrations-perforce", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=True)
     manager.add("organizations:scm-source-context", OrganizationFeature, FeatureHandlerStrategy.FLAGPOLE, api_expose=True)
     # Project Management Integrations Feature Parity Flags

src/sentry/integrations/github/tasks/__init__.py

@@ -2,12 +2,15 @@
 from .codecov_account_unlink import codecov_account_unlink
 from .link_all_repos import link_all_repos
 from .pr_comment import github_comment_workflow
+from .sync_repos import github_repo_sync_beat, sync_repos_for_org
 from .sync_repos_on_install_change import sync_repos_on_install_change
 
 __all__ = (
     "codecov_account_link",
     "codecov_account_unlink",
     "github_comment_workflow",
+    "github_repo_sync_beat",
     "link_all_repos",
+    "sync_repos_for_org",
     "sync_repos_on_install_change",
 )

src/sentry/integrations/github/tasks/sync_repos.py

@@ -0,0 +1,220 @@
+"""
+Periodic repo sync for GitHub integrations.
+
+The beat task (`github_repo_sync_beat`) runs on a schedule and uses
+CursoredScheduler to iterate over all active GitHub OrganizationIntegrations.
+For each one, it dispatches `sync_repos_for_org` which diffs GitHub's repo
+list against Sentry's Repository table and creates/disables/re-enables as needed.
+"""
+
+import logging
+from datetime import timedelta
+
+from taskbroker_client.retry import Retry
+
+from sentry import features
+from sentry.constants import ObjectStatus
+from sentry.integrations.models.organization_integration import OrganizationIntegration
+from sentry.integrations.services.integration import integration_service
+from sentry.integrations.services.repository.service import repository_service
+from sentry.integrations.source_code_management.metrics import (
+    SCMIntegrationInteractionEvent,
+    SCMIntegrationInteractionType,
+)
+from sentry.organizations.services.organization import organization_service
+from sentry.plugins.providers.integration_repository import (
+    RepoExistsError,
+    get_integration_repository_provider,
+)
+from sentry.shared_integrations.exceptions import ApiError
+from sentry.silo.base import SiloMode
+from sentry.tasks.base import instrumented_task, retry
+from sentry.taskworker.namespaces import integrations_control_tasks
+from sentry.utils import metrics
+from sentry.utils.cursored_scheduler import CursoredScheduler
+
+from .link_all_repos import get_repo_config
+
+logger = logging.getLogger(__name__)
+
+
+@instrumented_task(
+    name="sentry.integrations.github.tasks.sync_repos.sync_repos_for_org",
+    namespace=integrations_control_tasks,
+    retry=Retry(times=3, delay=120),
+    processing_deadline_duration=120,
+    silo_mode=SiloMode.CONTROL,
+)
+@retry(exclude=(RepoExistsError, KeyError))
+def sync_repos_for_org(organization_integration_id: int) -> None:
+    """
+    Sync repositories for a single OrganizationIntegration.
+
+    Fetches all repos from GitHub, diffs against Sentry's Repository table,
+    and creates/disables/re-enables repos as needed.
+    """
+    try:
+        oi = OrganizationIntegration.objects.get(
+            id=organization_integration_id,
+            status=ObjectStatus.ACTIVE,
+        )
+    except OrganizationIntegration.DoesNotExist:
+        logger.info(
+            "sync_repos_for_org.missing_org_integration",
+            extra={"organization_integration_id": organization_integration_id},
+        )
+        return
+
+    integration = integration_service.get_integration(
+        integration_id=oi.integration_id, status=ObjectStatus.ACTIVE
+    )
+    if integration is None:
+        logger.info(
+            "sync_repos_for_org.missing_integration",
+            extra={"integration_id": oi.integration_id},
+        )
+        return
+
+    organization_id = oi.organization_id
+    rpc_org = organization_service.get(id=organization_id)
+    if rpc_org is None:
+        logger.info(
+            "sync_repos_for_org.missing_organization",
+            extra={"organization_id": organization_id},
+        )
+        return
+
+    if not features.has("organizations:github-repo-auto-sync", rpc_org):
+        return
+
+    provider = f"integrations:{integration.provider}"
+    dry_run = not features.has("organizations:github-repo-auto-sync-apply", rpc_org)
+
+    with SCMIntegrationInteractionEvent(
+        interaction_type=SCMIntegrationInteractionType.SYNC_REPOS,
+        integration_id=integration.id,
+        organization_id=organization_id,
+        provider_key=integration.provider,
+    ).capture():
+        installation = integration.get_installation(organization_id=organization_id)
+        client = installation.get_client()
+
+        try:
+            github_repos = client.get_repos()
+        except ApiError as e:
+            if installation.is_rate_limited_error(e):
+                logger.info(
+                    "sync_repos_for_org.rate_limited",
+                    extra={
+                        "integration_id": integration.id,
+                        "organization_id": organization_id,
+                    },
+                )
+            raise
+
+        github_external_ids = {str(repo["id"]) for repo in github_repos}
+
+        all_repos = repository_service.get_repositories(
+            organization_id=organization_id,
+            integration_id=integration.id,
+            providers=[provider],
+        )
+        active_repos = [r for r in all_repos if r.status == ObjectStatus.ACTIVE and r.external_id]
+        disabled_repos = [
+            r for r in all_repos if r.status == ObjectStatus.DISABLED and r.external_id
+        ]
+
+        sentry_active_ids = {r.external_id for r in active_repos}
+        sentry_disabled_ids = {r.external_id for r in disabled_repos}
+
+        new_ids = github_external_ids - sentry_active_ids - sentry_disabled_ids
+        removed_ids = sentry_active_ids - github_external_ids
+        restored_ids = sentry_disabled_ids & github_external_ids
+
+        metric_tags = {
+            "provider": integration.provider,
+            "dry_run": str(dry_run),
+        }
+        metrics.distribution("scm.repo_sync.new_repos", len(new_ids), tags=metric_tags)
+        metrics.distribution("scm.repo_sync.removed_repos", len(removed_ids), tags=metric_tags)
+        metrics.distribution("scm.repo_sync.restored_repos", len(restored_ids), tags=metric_tags)
+        metrics.distribution(
+            "scm.repo_sync.provider_total", len(github_external_ids), tags=metric_tags
+        )
+        metrics.distribution(
+            "scm.repo_sync.sentry_active", len(sentry_active_ids), tags=metric_tags
+        )
+        metrics.distribution(
+            "scm.repo_sync.sentry_disabled", len(sentry_disabled_ids), tags=metric_tags
+        )
+
+        if new_ids or removed_ids or restored_ids:
+            logger.info(
+                "scm.repo_sync.diff",
+                extra={
+                    "provider": integration.provider,
+                    "integration_id": integration.id,
+                    "organization_id": organization_id,
+                    "dry_run": dry_run,
+                    "provider_total": len(github_external_ids),
+                    "sentry_active": len(sentry_active_ids),
+                    "sentry_disabled": len(sentry_disabled_ids),
+                    "new": len(new_ids),
+                    "removed": len(removed_ids),
+                    "restored": len(restored_ids),
+                },
+            )
+
+        if dry_run:
+            return
+
+        if new_ids:
+            integration_repo_provider = get_integration_repository_provider(integration)
+            repo_configs = [
+                get_repo_config(repo, integration.id)
+                for repo in github_repos
+                if str(repo["id"]) in new_ids
+            ]
+            if repo_configs:
+                try:
+                    integration_repo_provider.create_repositories(
+                        configs=repo_configs, organization=rpc_org
+                    )
+                except RepoExistsError:
+                    pass
+
+        if removed_ids:
+            repository_service.disable_repositories_by_external_ids(
+                organization_id=organization_id,
+                integration_id=integration.id,
+                provider=provider,
+                external_ids=list(removed_ids),
+            )
+
+        if restored_ids:
+            for repo in disabled_repos:
+                if repo.external_id in restored_ids:
+                    repo.status = ObjectStatus.ACTIVE
+                    repository_service.update_repository(
+                        organization_id=organization_id, update=repo
+                    )
+
+
+@instrumented_task(
+    name="sentry.integrations.github.tasks.sync_repos.github_repo_sync_beat",
+    namespace=integrations_control_tasks,
+    silo_mode=SiloMode.CONTROL,
+)
+def github_repo_sync_beat() -> None:
+    scheduler = CursoredScheduler(
+        name="github_repo_sync",
+        schedule_key="github-repo-sync-beat",
+        queryset=OrganizationIntegration.objects.filter(
+            integration__provider="github",
+            integration__status=ObjectStatus.ACTIVE,
+            status=ObjectStatus.ACTIVE,
+        ),
+        task=sync_repos_for_org,
+        cycle_duration=timedelta(hours=24),
+    )
+    scheduler.tick()

src/sentry/integrations/source_code_management/metrics.py

@@ -41,6 +41,7 @@ class SCMIntegrationInteractionType(StrEnum):
 
     # Tasks
     LINK_ALL_REPOS = "link_all_repos"
+    SYNC_REPOS = "sync_repos"
     SYNC_REPOS_ON_INSTALL_CHANGE = "sync_repos_on_install_change"
 
     # GitHub only