[airflow] branch master updated: Fix error with quick-failing tasks in KubernetesPodOperator (#13621)

[di...@apache.org]

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

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


The following commit(s) were added to refs/heads/master by this push:
     new 94d3ed6  Fix error with quick-failing tasks in KubernetesPodOperator (#13621)
94d3ed6 is described below

commit 94d3ed61d60b134d649a4e9785b2d9c2a88cff05
Author: Daniel Imberman <da...@gmail.com>
AuthorDate: Thu Jan 21 12:57:35 2021 -0800

    Fix error with quick-failing tasks in KubernetesPodOperator (#13621)
    
    * Fix error with quick-failing tasks in KubernetesPodOperator
    
    Addresses an issue with the KubernetesPodOperator where tasks that die
    quickly are not patched with "already_checked" because they never make
    it to the monitoring logic.
    
    * static fix
---
 .../cncf/kubernetes/operators/kubernetes_pod.py    |  1 +
 kubernetes_tests/test_kubernetes_pod_operator.py   | 25 ++++++++++++++++++++++
 2 files changed, 26 insertions(+)

diff --git a/airflow/providers/cncf/kubernetes/operators/kubernetes_pod.py b/airflow/providers/cncf/kubernetes/operators/kubernetes_pod.py
index 3f42ab1..996b7da 100644
--- a/airflow/providers/cncf/kubernetes/operators/kubernetes_pod.py
+++ b/airflow/providers/cncf/kubernetes/operators/kubernetes_pod.py
@@ -343,6 +343,7 @@ class KubernetesPodOperator(BaseOperator):  # pylint: disable=too-many-instance-
                 raise AirflowException(f'Pod {self.pod.metadata.name} returned a failure: {status}')
             return result
         except AirflowException as ex:
+            self.patch_already_checked(self.pod)
             raise AirflowException(f'Pod Launching failed: {ex}')
 
     def handle_pod_overlap(
diff --git a/kubernetes_tests/test_kubernetes_pod_operator.py b/kubernetes_tests/test_kubernetes_pod_operator.py
index 574f7ff..e860303 100644
--- a/kubernetes_tests/test_kubernetes_pod_operator.py
+++ b/kubernetes_tests/test_kubernetes_pod_operator.py
@@ -1000,5 +1000,30 @@ class TestKubernetesPodOperatorSystem(unittest.TestCase):
             k.execute(context)
             create_mock.assert_called_once()
 
+    def test_reatttach_quick_failure(self):
+        client = kube_client.get_kube_client(in_cluster=False)
+        namespace = "default"
+
+        name = "test"
+        k = KubernetesPodOperator(
+            namespace='default',
+            image="ubuntu:16.04",
+            cmds=["bash", "-cx"],
+            arguments=["exit 1"],
+            labels={"foo": "bar"},
+            name="test",
+            task_id=name,
+            in_cluster=False,
+            do_xcom_push=False,
+            is_delete_operator_pod=False,
+            termination_grace_period=0,
+        )
+
+        context = create_context(k)
+        with self.assertRaises(AirflowException):
+            k.execute(context)
+        pod = client.read_namespaced_pod(name=k.pod.metadata.name, namespace=namespace)
+        self.assertEqual(pod.metadata.labels["already_checked"], "True")
+
 
 # pylint: enable=unused-argument