amoghrajesh commented on code in PR #67715:
URL: https://github.com/apache/airflow/pull/67715#discussion_r3327980908
##########
providers/apache/spark/src/airflow/providers/apache/spark/hooks/spark_submit.py:
##########
@@ -682,10 +730,17 @@ def _process_spark_submit_log(self, itr: Iterator[Any])
-> None:
# If we run Kubernetes cluster mode, we want to extract the driver
pod id
# from the logs so we can kill the application when we stop it
unexpectedly
elif self._is_kubernetes:
+ # Two log formats exist across Spark versions:
+ # "pod name: <name>-driver" and "submission ID
spark:<name>-driver"
match_driver_pod = re.search(r"\s*pod name:
((.+?)-([a-z0-9]+)-driver$)", line)
if match_driver_pod:
self._kubernetes_driver_pod = match_driver_pod.group(1)
self.log.info("Identified spark driver pod: %s",
self._kubernetes_driver_pod)
+ if not self._kubernetes_driver_pod:
+ match_submission_id = re.search(r"submission ID
spark:(.+-driver)", line)
Review Comment:
Handled it in: e42f28a089
##########
providers/apache/spark/src/airflow/providers/apache/spark/hooks/spark_submit.py:
##########
@@ -802,6 +857,54 @@ def _start_driver_status_tracking(self) -> None:
f"returncode = {returncode}"
)
+ def _poll_k8s_driver_via_api(self) -> None:
+ """Poll the K8s driver pod phase until it reaches a terminal state."""
+ pod_name = self._kubernetes_driver_pod
+ namespace = self._connection["namespace"]
+ app_id = self._kubernetes_application_id or pod_name
+
+ if not pod_name:
+ raise ValueError("K8s driver pod name not set; cannot poll
status.")
+
+ client = kube_client.get_kube_client(in_cluster=False)
+ poll_interval = max(self._status_poll_interval, 20)
Review Comment:
Handled it in: e42f28a089
##########
providers/apache/spark/src/airflow/providers/apache/spark/hooks/spark_submit.py:
##########
@@ -802,6 +857,54 @@ def _start_driver_status_tracking(self) -> None:
f"returncode = {returncode}"
)
+ def _poll_k8s_driver_via_api(self) -> None:
+ """Poll the K8s driver pod phase until it reaches a terminal state."""
+ pod_name = self._kubernetes_driver_pod
+ namespace = self._connection["namespace"]
Review Comment:
Handled it in: e42f28a089
##########
providers/apache/spark/src/airflow/providers/apache/spark/hooks/spark_submit.py:
##########
@@ -41,6 +41,9 @@
DEFAULT_SPARK_BINARY = "spark-submit"
ALLOWED_SPARK_BINARIES = [DEFAULT_SPARK_BINARY, "spark2-submit",
"spark3-submit"]
+_K8S_WAIT_APP_COMPLETION_CONF = "spark.kubernetes.submission.waitAppCompletion"
+_K8S_DELETE_ON_TERMINATION_CONF = "spark.kubernetes.driver.deleteOnTermination"
Review Comment:
Handled it in: e42f28a089
##########
providers/apache/spark/src/airflow/providers/apache/spark/operators/spark_submit.py:
##########
@@ -234,6 +242,13 @@ def execute(self, context: Context) -> None:
driver_id = self.submit_job(context)
self.poll_until_complete(driver_id, context)
return self.get_job_result(driver_id, context)
+ if hook._should_track_driver_via_k8s_api():
+ hook._validate_track_driver_via_k8s_api_config()
Review Comment:
Handled it in: e42f28a089
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
