[
https://issues.apache.org/jira/browse/FLINK-37831?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Thomas Bay updated FLINK-37831:
-------------------------------
Description:
When using flink-connector-elasticsearch8 ver. 3.1.0-1.20 with flink ver.
1.20.1,
the retry of failed index operations has two problems in
Elasticsearch8AsyncWriter class:
* the isRetryable method is testing error.getCause() which is null in case of
ConnectException and other exceptions
* ELASTICSEARCH_FATAL_EXCEPTION_CLASSIFIER.isFatal() method returns true if
exception is not thrown, resulting in the isRetryable() method returns the
wrong result.
The fix for ver. 3.1. is included in [Pull Request
127|https://github.com/apache/flink-connector-elasticsearch/pull/127]
The committed fixed the problem for me.
was:
When using flink-connector-elasticsearch8 ver. 3.1.0-1.20 with flink ver.
1.20.1,
the retry of failed index operations has two problems:
* the isRetryable method is testing error.getCause() which is null in case of
ConnectException and other exceptions
* ELASTICSEARCH_FATAL_EXCEPTION_CLASSIFIER.isFatal() method returns true if
exception is not thrown, resulting in the isRetryable() method returns the
wrong result.
The fix for ver. 3.1. is included in [Pull Request
127|https://github.com/apache/flink-connector-elasticsearch/pull/127]
The committed fixed the problem for me.
> Failed index operations are not retried when using
> flink-connector-elasticsearch8 ver. 3.1.0-1.20 with flink ver. 1.20.1
> ------------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-37831
> URL: https://issues.apache.org/jira/browse/FLINK-37831
> Project: Flink
> Issue Type: Bug
> Components: Connectors / ElasticSearch
> Environment: flink-connector-elasticsearch8 ver. 3.1.0-1.20
> flink ver. 1.20.1,
> Reporter: Thomas Bay
> Priority: Major
> Labels: pull-request-available
>
> When using flink-connector-elasticsearch8 ver. 3.1.0-1.20 with flink ver.
> 1.20.1,
> the retry of failed index operations has two problems in
> Elasticsearch8AsyncWriter class:
> * the isRetryable method is testing error.getCause() which is null in case of
> ConnectException and other exceptions
> * ELASTICSEARCH_FATAL_EXCEPTION_CLASSIFIER.isFatal() method returns true if
> exception is not thrown, resulting in the isRetryable() method returns the
> wrong result.
> The fix for ver. 3.1. is included in [Pull Request
> 127|https://github.com/apache/flink-connector-elasticsearch/pull/127]
> The committed fixed the problem for me.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
