Martijn Visser created FLINK-41045:
--------------------------------------
Summary: unning Kerberized YARN application on Docker test (custom
fs plugin)' fails with NoResourceAvailableException because two of three
TaskManager containers never register
Key: FLINK-41045
URL: https://issues.apache.org/jira/browse/FLINK-41045
Project: Flink
Issue Type: Bug
Components: Deployment / YARN
Affects Versions: 1.20.6
Reporter: Martijn Visser
https://dev.azure.com/apache-flink/98463496-1af2-4620-8eab-a2ecc1a2e6fe/_build/results?buildId=80244&view=logs&j=bbb1e2a2-a43c-55c8-fb48-5cfe7a8a0ca6
(release-1.20 nightly 2026-10-18, e2e_1_cron_hadoop313)
The application requested three TaskManager containers at 03:05:57. The one on
worker1 registered at 03:06:12; the two on worker2 never did, and the job
failed when the 120 s slot request timeout fired:
{code}
2026-10-18 03:06:12,324 INFO
org.apache.flink.runtime.resourcemanager.active.ActiveResourceManager [] -
Worker
container_1792292630021_0001_01_000002(worker1.docker-hadoop-cluster-network:43305)
is registered.
...
org.apache.flink.runtime.jobmanager.scheduler.NoResourceAvailableException:
Slot request bulk is not fulfillable! Could not allocate the required slot
within slot request timeout
Caused by: java.util.concurrent.TimeoutException: Timeout has occurred: 120000
ms
...
Final Application State: FAILED
[FAIL] 'Running Kerberized YARN application on Docker test (custom fs plugin)'
failed after 5 minutes and 17 seconds! Test exited with exit code 1
{code}
The worker2 container localizer was still reading from HDFS when the
application shut down. FLINK-22566 was the same 120 s timeout in 2021 and was
closed as Cannot Reproduce. The test runs again on release-1.20 since
FLINK-40661.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
