[
https://issues.apache.org/jira/browse/FLINK-20292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17237832#comment-17237832
]
Yun Gao commented on FLINK-20292:
---------------------------------
Very thanks [~lzljs3620320] for the explanation! And adding the transformer to
the shaded plugin configuration did solve this issue:
{code:java}
<!-- The service transformer is needed to merge META-INF/services files -->
<transformer
implementation="org.apache.maven.plugins.shade.resource.ServicesResourceTransformer"/>
{code}
I agree with that we could improve the document to alert users. Also, I saw
that Flink's project architecture _flink-walkthrough-datastream-java_ also do
not have this configuration, it might be also better to add the configuration
in the architecture.
> Avro format conflicts with Parquet and Orc
> ------------------------------------------
>
> Key: FLINK-20292
> URL: https://issues.apache.org/jira/browse/FLINK-20292
> Project: Flink
> Issue Type: Bug
> Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile)
> Reporter: Yun Gao
> Assignee: Leonard Xu
> Priority: Critical
> Fix For: 1.12.0
>
>
> When testing the Flink 1.12 in a standalone mode cluster, I found that if the
> user job jar contains both _flink-avro_ and _flink-parquet/flink-orc_, the
> FileSystemTableSink would not be able to load the corresponding format
> factory correctly. But if only one format is dependent it works.
> The test project located in
> [here|https://github.com/gaoyunhaii/flink1.12test] and the test class is
> [FileCompactionTest|https://github.com/gaoyunhaii/flink1.12test/blob/main/src/main/java/FileCompactionTest.java].
> The conflict does not seem to affect the local runner, but only has problem
> when submitted to the standalone cluster.
> If the problem does exists, we might need to fix it or give user some tips
> about the conflicts.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
