You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@apex.apache.org by "Sandesh (JIRA)" <ji...@apache.org> on 2017/03/02 18:26:46 UTC
[jira] [Updated] (APEXMALHAR-2422) WindowDataManager not recovering
as expected on HDFS
[ https://issues.apache.org/jira/browse/APEXMALHAR-2422?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sandesh updated APEXMALHAR-2422:
--------------------------------
Description:
In WindowDataManager getLargestWindow during recovery is not reflecting the saved window before operator crash.
This affects many of the operators, one example is KafkaExactlyOnceOutputOperator, https://issues.apache.org/jira/browse/APEXMALHAR-2419
Note:
WindowDataManager works as expected on a local machine(local filesystem), but not on HDFS.
Testing for the fix
1. Updated the unit test to use Hadoop filesystem APIs and ran them on HDFS
2. Wrote a simple application to test the behavior of Window Data Manager on the cluster
was:
In WindowDataManager getLargestWindow during recovery is not reflecting the saved window before operator crash.
This affects many of the operators, one example is KafkaExactlyOnceOutputOperator, https://issues.apache.org/jira/browse/APEXMALHAR-2419
Note:
WindowDataManager works as expected on a local machine(local filesystem), but not on HDFS.
> WindowDataManager not recovering as expected on HDFS
> ----------------------------------------------------
>
> Key: APEXMALHAR-2422
> URL: https://issues.apache.org/jira/browse/APEXMALHAR-2422
> Project: Apache Apex Malhar
> Issue Type: Bug
> Reporter: Sandesh
> Assignee: Sandesh
>
> In WindowDataManager getLargestWindow during recovery is not reflecting the saved window before operator crash.
> This affects many of the operators, one example is KafkaExactlyOnceOutputOperator, https://issues.apache.org/jira/browse/APEXMALHAR-2419
> Note:
> WindowDataManager works as expected on a local machine(local filesystem), but not on HDFS.
> Testing for the fix
> 1. Updated the unit test to use Hadoop filesystem APIs and ran them on HDFS
> 2. Wrote a simple application to test the behavior of Window Data Manager on the cluster
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)