Dale Lane created FLINK-40860:
---------------------------------
Summary: Add bin/minicluster.sh to run an application on a
MiniCluster
Key: FLINK-40860
URL: https://issues.apache.org/jira/browse/FLINK-40860
Project: Flink
Issue Type: Sub-task
Components: Deployment / Scripts, Runtime / Coordination
Reporter: Dale Lane
Flink's MiniCluster runs a real JobManager and one or more TaskManagers merged
into a single JVM, instantiating the same component classes as a distributed
deployment. It is fully featured when used programmatically, but there is no
operational way to run a job on one.
{{flink run -t local}} does not offer:
* a job identity that is stable across process restarts, so checkpoint lineage
survives
* useful behaviour when HA metadata already contains the job, because a
previous process instance was terminated
* a process lifecycle suitable for a container runtime
Achieving this today requires writing Java against MiniCluster's internal APIs.
This issue adds {{bin/minicluster.sh}} and a supporting entrypoint class in
flink-container, giving MiniCluster the same kind of bootstrap that JobManager,
TaskManager and standalone-job already have.
This is additive: no changes to existing interfaces, scripts or behaviours, and
{{flink run -t local}} is unaffected.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
