Spongebob created FLINK-21279:
---------------------------------
Summary: flink on yarn sink nothing
Key: FLINK-21279
URL: https://issues.apache.org/jira/browse/FLINK-21279
Project: Flink
Issue Type: Bug
Components: Table SQL / API
Affects Versions: 1.12.1
Environment: flink: 1.12.1
hive: 3.1.2
Reporter: Spongebob
Attachments: yarn.log
Here's the data chain of the flink application:
# read from HDFS file via ExecutionEnvironment, get Dataset
# collect the Dataset into Seq object.
# transform Seq object into multi ArrayBuffer[Expression] objects.
# create TableObjects from ArrayBuffers using `fromValues`
# create catalog views from TableObjects
# sink into hive table from catalog views.
It is all normal until step 6. And it runs inconsistently on local IDE and yarn
cluster. When I run the application on local IDE it cost all network buffer
memory then turn into failed( actually the HDFS file size is less then 2MB, and
I had set the parrallelism of tableEnv to 1. If I run the one sink of them only
it can be run normally). And to the yarn cluster, there throws the exception
`Job was submitted in detached mode. Results of job execution, such as
accumulators, runtime, etc. are not available. Please make sure your program
doesn't call an eager execution function [collect, print, printToErr, count]`
but the application can run successfully however can not sink anything to hive,
I find it does not request any slot while running.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
