[jira] [Commented] (FLINK-9143) Restart strategy defined in flink-conf.yaml is ignored

["ASF GitHub Bot (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-9143?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16487057#comment-16487057 ] 

ASF GitHub Bot commented on FLINK-9143:
---------------------------------------

Github user dawidwys commented on the issue:

    https://github.com/apache/flink/pull/5846
  
    @yuqi1129 Thank you for your contribution. 
    
    Did I understand this change correctly that it only affects situation when the checkpointing is disabled? Currently in this case the `RestartStrategy` will be set from cluster's `flink-conf.yaml`. After your change the cluster configuration will never be taken into account (neither with nor without checkpointing). Therefore I would be against merging it in current shape, as it does not address the original problem, but just changes only one of many corner cases.
    
    What do you think about such strategy for resolving `RestartStrategy`?:
    
    1. job config
      * set in code
      * client configuration
      * (no default, even in case of checkpointing)
    2. cluster config
      * cluster configuration (flink-conf)
      * default value = FixedDelayRestart (might be only in case of checkpointing)


> Restart strategy defined in flink-conf.yaml is ignored
> ------------------------------------------------------
>
>                 Key: FLINK-9143
>                 URL: https://issues.apache.org/jira/browse/FLINK-9143
>             Project: Flink
>          Issue Type: Bug
>          Components: Configuration
>    Affects Versions: 1.4.2
>            Reporter: Alex Smirnov
>            Assignee: yuqi
>            Priority: Major
>         Attachments: execution_config.png, jobmanager.log, jobmanager.png
>
>
> Restart strategy defined in flink-conf.yaml is disregarded, when user enables checkpointing.
> Steps to reproduce:
> 1. Download flink distribution (1.4.2), update flink-conf.yaml:
>   
>  restart-strategy: none
>  state.backend: rocksdb
>  state.backend.fs.checkpointdir: [file:///tmp/nfsrecovery/flink-checkpoints-metadata]
>  state.backend.rocksdb.checkpointdir: [file:///tmp/nfsrecovery/flink-checkpoints-rocksdb]
>   
>  2. create new java project as described at [https://ci.apache.org/projects/flink/flink-docs-release-1.4/quickstart/java_api_quickstart.html]
>  here's the code:
>  public class FailedJob
>  {
>      static final Logger LOGGER = LoggerFactory.getLogger(FailedJob.class);
>      public static void main( String[] args ) throws Exception
>      {
>          final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
>          env.enableCheckpointing(5000, CheckpointingMode.EXACTLY_ONCE);
>          DataStream<String> stream = env.fromCollection(Arrays.asList("test"));
>          stream.map(new MapFunction<String, String>(){
>              @Override
>              public String map(String obj)
> {                 throw new NullPointerException("NPE");             }
>  
>          });
>          env.execute("Failed job");
>      }
>  }
>   
>  3. Compile: mvn clean package; submit it to the cluster
>   
>  4. Go to Job Manager configuration in WebUI, ensure settings from flink-conf.yaml is there (screenshot attached)
>   
>  5. Go to Job's configuration, see Execution Configuration section
>   
>  *Expected result*: restart strategy as defined in flink-conf.yaml
>   
>  *Actual result*: Restart with fixed delay (10000 ms). #2147483647 restart attempts.
>   
>   
>  see attached screenshots and jobmanager log (line 1 and 31)
>   



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)