[
https://issues.apache.org/jira/browse/FLINK-6296?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Petr Novotnik updated FLINK-6296:
---------------------------------
Description:
Hello,
calling {{ExecutionEnvironment#getExecutionPlan()}} prior to
{{ExecutionEnvironment#execute()}} makes the later fail with the following
message if I try to run my job on yarn, i.e. {{flink run -m yarn-cluster ...}}:
{noformat}
> Caused by: java.lang.RuntimeException: No new data sinks have been defined
> since the last execution. The last execution refers to the latest call to
> 'execute()', 'count()', 'collect()', or 'print()'.
> at
> org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1050)
> at
> org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1032)
> at
> org.apache.flink.client.program.ContextEnvironment.execute(ContextEnvironment.java:59)
> at
> org.apache.flink.api.java.ExecutionEnvironment.execute(ExecutionEnvironment.java:926)
{noformat}
It works fine when executing the job locally (and it works fine locally or even
on yarn using the DataStream API, i.e. {{StreamExecutionEnvironment}}).
So far I could track it down, the problem seems the "clearing" behaviour of
{{ContextEnvironment#createProgramPlan(..)}}. This makes me wonder why
{{#getExecutionPlan}} uses the one parameter version of
{{#createProgramPlan(..)}}. Is "clearing" the sinks really required as part of
{{getExecutionPlan()}}?
>From a [code
>perspective](https://github.com/apache/flink/blob/release-1.1.5/flink-clients/src/main/java/org/apache/flink/client/program/ContextEnvironment.java#L68),
> it seems the same is present on Flink 1.1.5, but I haven't tested it,
P.
was:
Hello,
calling {{ExecutionEnvironment#getExecutionPlan()}} prior to
{{ExecutionEnvironment#execute()}} makes the later fail with the following
message if I try to run my job on yarn, i.e. {{flink run -m yarn-cluster ...}}:
{noformat}
> Caused by: java.lang.RuntimeException: No new data sinks have been defined
> since the last execution. The last execution refers to the latest call to
> 'execute()', 'count()', 'collect()', or 'print()'.
> at
> org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1050)
> at
> org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1032)
> at
> org.apache.flink.client.program.ContextEnvironment.execute(ContextEnvironment.java:59)
> at
> org.apache.flink.api.java.ExecutionEnvironment.execute(ExecutionEnvironment.java:926)
{noformat}
It works fine when executing the job locally (and it works fine locally or even
on yarn using the DataStream API, i.e. {{StreamExecutionEnvironment}}).
So far I could track it down, the problem seems the "clearing" behaviour of
{{ContextEnvironment#createProgramPlan(..)}}. This makes me wonder why
{{#getExecutionPlan}} uses the one parameter version of
{{#createProgramPlan(..)}}. Is "clearing" the sinks really required as part of
{{getExecutionPlan()}}?
I just check, in Flink 1.1.5, {{#getExecutionPlan(..)}} invokes
{{#createProgramPlan("unnamed job", false)}}, hence, it would be good to
document as a breaking change [in the migration
guide|https://ci.apache.org/projects/flink/flink-docs-release-1.2/ops/upgrading.html]
since the resulting error message is rather misleading.
P.
> Retrieving execution plan fails succeeding execution attempt
> ------------------------------------------------------------
>
> Key: FLINK-6296
> URL: https://issues.apache.org/jira/browse/FLINK-6296
> Project: Flink
> Issue Type: Bug
> Components: DataSet API, Job-Submission, YARN
> Affects Versions: 1.2.0
> Reporter: Petr Novotnik
>
> Hello,
> calling {{ExecutionEnvironment#getExecutionPlan()}} prior to
> {{ExecutionEnvironment#execute()}} makes the later fail with the following
> message if I try to run my job on yarn, i.e. {{flink run -m yarn-cluster
> ...}}:
> {noformat}
> > Caused by: java.lang.RuntimeException: No new data sinks have been defined
> > since the last execution. The last execution refers to the latest call to
> > 'execute()', 'count()', 'collect()', or 'print()'.
> > at
> > org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1050)
> > at
> > org.apache.flink.api.java.ExecutionEnvironment.createProgramPlan(ExecutionEnvironment.java:1032)
> > at
> > org.apache.flink.client.program.ContextEnvironment.execute(ContextEnvironment.java:59)
> > at
> > org.apache.flink.api.java.ExecutionEnvironment.execute(ExecutionEnvironment.java:926)
> {noformat}
> It works fine when executing the job locally (and it works fine locally or
> even on yarn using the DataStream API, i.e. {{StreamExecutionEnvironment}}).
> So far I could track it down, the problem seems the "clearing" behaviour of
> {{ContextEnvironment#createProgramPlan(..)}}. This makes me wonder why
> {{#getExecutionPlan}} uses the one parameter version of
> {{#createProgramPlan(..)}}. Is "clearing" the sinks really required as part
> of {{getExecutionPlan()}}?
> From a [code
> perspective](https://github.com/apache/flink/blob/release-1.1.5/flink-clients/src/main/java/org/apache/flink/client/program/ContextEnvironment.java#L68),
> it seems the same is present on Flink 1.1.5, but I haven't tested it,
> P.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
