tree 56d7b24d9125eb81d1e8d763fd3ffb5e8419bd4b
parent 85314ddbc70839913f16121ae60293c6c05f0a76
author haomarch <marchpure@126.com> 1596697278 +0800
committer QiangCai <qiangcai@qq.com> 1597023893 +0800

[CARBONDATA-3940] Remove "spark.sql.sources.commitProtocolClass" configuration from SparkCarbonTableFormat

Why is this PR needed?
During the load process, commitTask fails with high probability.
The exception stack shows that it was throwed by HadoopMapReduceCommitProtocol, not CarbonSQLHadoopMapMapReduceCommitProtocol,
implying that there is class init error during the initializing of "Committer".
It should have been initialized as CarbonSQLHadoopMapMapReduceCommitProtocol, but was incorrectly initialized to HadoopMapReduceCommitProtocol.

What changes were proposed in this PR?
Init the committer to be CarbonSQLHadoopMapMapReduceCommitProtocol directly

Does this PR introduce any user interface change?
No

Is any new testcase added?
No

This closes #3883