(airflow) branch main updated: Fix Git corruption recovery by moving fetch into retry context (#56913)

[potiuk]

This is an automated email from the ASF dual-hosted git repository.

potiuk pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/airflow.git


The following commit(s) were added to refs/heads/main by this push:
     new b32603f5066 Fix Git corruption recovery by moving fetch into retry 
context (#56913)
b32603f5066 is described below

commit b32603f50668391c2bba55579336023fab6dbf0f
Author: Dheeraj Turaga <dheerajtur...@gmail.com>
AuthorDate: Mon Oct 27 19:56:14 2025 -0500

    Fix Git corruption recovery by moving fetch into retry context (#56913)
    
    * Fix Git corruption recovery by moving fetch into retry context
    
      Move _fetch_bare_repo call from _ensure_version_in_bare_repo into
      _clone_bare_repo_if_required so that fetch failures (e.g., corrupted packs
      with unresolved deltas) trigger the existing @retry mechanism to clean up
      and re-clone the bare repository automatically.
    
    * Fix tests
---
 providers/git/src/airflow/providers/git/bundles/git.py | 5 ++++-
 providers/git/tests/unit/git/bundles/test_git.py       | 7 +++++--
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/providers/git/src/airflow/providers/git/bundles/git.py 
b/providers/git/src/airflow/providers/git/bundles/git.py
index 89c7f7af084..2bf1f9980ae 100644
--- a/providers/git/src/airflow/providers/git/bundles/git.py
+++ b/providers/git/src/airflow/providers/git/bundles/git.py
@@ -175,9 +175,12 @@ class GitDagBundle(BaseDagBundle):
                     env=self.hook.env if self.hook else None,
                 )
             self.bare_repo = Repo(self.bare_repo_path)
+
+            # Fetch to ensure we have latest refs and validate repo integrity
+            self._fetch_bare_repo()
         except (InvalidGitRepositoryError, GitCommandError) as e:
             self._log.warning(
-                "Bare repository clone/open failed, cleaning up and retrying",
+                "Bare repository clone/open/fetch failed, cleaning up and 
retrying",
                 bare_repo_path=self.bare_repo_path,
                 exc=e,
             )
diff --git a/providers/git/tests/unit/git/bundles/test_git.py 
b/providers/git/tests/unit/git/bundles/test_git.py
index cf85e71758b..b6c110e3547 100644
--- a/providers/git/tests/unit/git/bundles/test_git.py
+++ b/providers/git/tests/unit/git/bundles/test_git.py
@@ -377,7 +377,8 @@ class TestGitDagBundle:
             tracking_ref=GIT_DEFAULT_BRANCH,
         )
         bundle.initialize()
-        assert mock_gitRepo.return_value.remotes.origin.fetch.call_count == 2  
# 1 in bare, 1 in main repo
+        # 1 in _clone_bare_repo_if_required, 1 in refresh() for bare repo, 1 
in refresh() for working repo
+        assert mock_gitRepo.return_value.remotes.origin.fetch.call_count == 3
         mock_gitRepo.return_value.remotes.origin.fetch.reset_mock()
         bundle.refresh()
         assert mock_gitRepo.return_value.remotes.origin.fetch.call_count == 2
@@ -730,7 +731,9 @@ class TestGitDagBundle:
         EXPECTED_ENV = {"GIT_SSH_COMMAND": "ssh -i /id_rsa -o 
StrictHostKeyChecking=no"}
 
         mock_gitRepo.clone_from.side_effect = _fake_clone_from
-        mock_gitRepo.return_value = types.SimpleNamespace()
+        # Mock needs to support the fetch operation called in 
_clone_bare_repo_if_required
+        mock_repo_instance = mock.MagicMock()
+        mock_gitRepo.return_value = mock_repo_instance
 
         with mock.patch("airflow.providers.git.bundles.git.GitHook") as 
mock_githook:
             mock_githook.return_value.repo_url = 
"g...@github.com:apache/airflow.git"