This is an automated email from the ASF dual-hosted git repository.
vincbeck 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 8637df49fcb Increase Comprehend document classifier timeout from 20 to
60 minutes (#71969)
8637df49fcb is described below
commit 8637df49fcb4a775422e14fc1fbee3a7ef62fc0e
Author: Ramit Kataria <[email protected]>
AuthorDate: Wed Aug 26 06:52:06 2026 -0700
Increase Comprehend document classifier timeout from 20 to 60 minutes
(#71969)
Classifier training sometimes takes longer than 20 minutes, but the
operator's default waiter budget was 20 x 60s in both sync and deferrable
paths, so those runs failed with a waiter timeout. Raise the cap to
60 x 60s: the companion sensor and trigger allow 150 minutes, but that is
too large a jump for the operator's implicit wait, so they are left
unchanged. This also lets the system test Dag drop its
wait_for_completion=False workaround, so the test exercises the operator's
default wait path. A changelog warning documents the longer default wait.
---
providers/amazon/docs/changelog.rst | 8 ++++++++
.../src/airflow/providers/amazon/aws/operators/comprehend.py | 4 ++--
.../system/amazon/aws/example_comprehend_document_classifier.py | 1 -
.../amazon/tests/unit/amazon/aws/operators/test_comprehend.py | 2 ++
4 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/providers/amazon/docs/changelog.rst
b/providers/amazon/docs/changelog.rst
index d7e44bbe5de..ca7ffc5acaa 100644
--- a/providers/amazon/docs/changelog.rst
+++ b/providers/amazon/docs/changelog.rst
@@ -26,6 +26,14 @@
Changelog
---------
+.. warning::
+ The default waiter timeout of ``ComprehendCreateDocumentClassifierOperator``
was raised from
+ 20 minutes (``waiter_max_attempts=20``) to 60 minutes
(``waiter_max_attempts=60``), because
+ document classifier training sometimes takes longer than 20 minutes. When
the operator waits
+ for completion (the default, in both synchronous and deferrable mode), tasks
that previously
+ failed with a waiter timeout around the 20-minute mark now keep waiting for
up to an hour.
+ Pass ``waiter_max_attempts`` explicitly to restore the previous timeout.
+
9.35.1
......
diff --git
a/providers/amazon/src/airflow/providers/amazon/aws/operators/comprehend.py
b/providers/amazon/src/airflow/providers/amazon/aws/operators/comprehend.py
index 575263eb0f6..cdc3a30bf9e 100644
--- a/providers/amazon/src/airflow/providers/amazon/aws/operators/comprehend.py
+++ b/providers/amazon/src/airflow/providers/amazon/aws/operators/comprehend.py
@@ -238,7 +238,7 @@ class
ComprehendCreateDocumentClassifierOperator(AwsBaseOperator[ComprehendHook]
:param wait_for_completion: Whether to wait for job to stop. (default:
True)
:param waiter_delay: Time in seconds to wait between status checks.
(default: 60)
- :param waiter_max_attempts: Maximum number of attempts to check for job
completion. (default: 20)
+ :param waiter_max_attempts: Maximum number of attempts to check for job
completion. (default: 60)
:param deferrable: If True, the operator will wait asynchronously for the
job to stop.
This implies waiting for completion. This mode requires aiobotocore
module to be installed.
(default: False)
@@ -285,7 +285,7 @@ class
ComprehendCreateDocumentClassifierOperator(AwsBaseOperator[ComprehendHook]
document_classifier_kwargs: dict[str, Any] | None = None,
wait_for_completion: bool = True,
waiter_delay: int = 60,
- waiter_max_attempts: int = 20,
+ waiter_max_attempts: int = 60,
deferrable: bool = conf.getboolean("operators", "default_deferrable",
fallback=False),
**kwargs,
):
diff --git
a/providers/amazon/tests/system/amazon/aws/example_comprehend_document_classifier.py
b/providers/amazon/tests/system/amazon/aws/example_comprehend_document_classifier.py
index 748db647eec..7b32eae9615 100644
---
a/providers/amazon/tests/system/amazon/aws/example_comprehend_document_classifier.py
+++
b/providers/amazon/tests/system/amazon/aws/example_comprehend_document_classifier.py
@@ -106,7 +106,6 @@ def document_classifier_workflow():
document_classifier_kwargs=document_classifier_kwargs,
)
# [END howto_operator_create_document_classifier]
- create_document_classifier.wait_for_completion = False
# [START howto_sensor_create_document_classifier]
await_create_document_classifier =
ComprehendCreateDocumentClassifierCompletedSensor(
diff --git
a/providers/amazon/tests/unit/amazon/aws/operators/test_comprehend.py
b/providers/amazon/tests/unit/amazon/aws/operators/test_comprehend.py
index fea16411a29..c9ed74b9e5d 100644
--- a/providers/amazon/tests/unit/amazon/aws/operators/test_comprehend.py
+++ b/providers/amazon/tests/unit/amazon/aws/operators/test_comprehend.py
@@ -243,6 +243,8 @@ class TestComprehendCreateDocumentClassifierOperator:
assert self.operator.language_code == "en"
assert self.operator.document_classifier_kwargs == {"VersionName":
"v1"}
assert self.operator.fail_on_warnings is False
+ assert self.operator.waiter_delay == 60
+ assert self.operator.waiter_max_attempts == 60
@mock.patch.object(ComprehendHook, "conn")
def test_create_document_classifier(self, mock_conn):
