[
https://issues.apache.org/jira/browse/FLINK-18085?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17126498#comment-17126498
]
Yang Wang edited comment on FLINK-18085 at 6/8/20, 8:33 AM:
------------------------------------------------------------
Test application mode for native K8s:
* Submit a Flink application cluster
* Submit a Flink application cluster with ZK HA configured
** Kill jobmanager, job could recover from latest checkpoint
* Cancel the only job, cluster should be destroyed automatically.
* [*FAILED*] Check the webui and logs
** The TaskManager logs could not show up.
Some tests failed. Will create a new ticket to track.
Update:
FLINK-18149 has been merged. The taskmanager logs could show up correctly.
was (Author: fly_in_gis):
Test application mode for native K8s:
* Submit a Flink application cluster
* Submit a Flink application cluster with ZK HA configured
** Kill jobmanager, job could recover from latest checkpoint
* Cancel the only job, cluster should be destroyed automatically.
* [*FAILED*] Check the webui and logs
** The TaskManager logs could not show up.
Some tests failed. Will create a new ticket to track.
> Manually test application mode for standalone, Yarn, K8s
> --------------------------------------------------------
>
> Key: FLINK-18085
> URL: https://issues.apache.org/jira/browse/FLINK-18085
> Project: Flink
> Issue Type: Sub-task
> Reporter: Yang Wang
> Assignee: Yang Wang
> Priority: Critical
> Labels: release-testing
> Fix For: 1.11.0
>
>
> We have introduced the application mode from 1.11. It need to be tested in a
> real cluster with following functionality check.
> * Application normally start and finish
> * Application failed exceptionally
> * HA configured, kill jobmanager and job should recover from latest
> checkpoint
>
> For Yarn deployment, we also need to verify it could work with provided lib
> and remote user jars.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
