moomindani commented on a change in pull request #27690:
URL: https://github.com/apache/spark/pull/27690#discussion_r439913383
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
##########
@@ -839,6 +839,17 @@ object SQLConf {
.checkValues(HiveCaseSensitiveInferenceMode.values.map(_.toString))
.createWithDefault(HiveCaseSensitiveInferenceMode.NEVER_INFER.toString)
+ val HIVE_SUPPORTED_SCHEMES_TO_USE_NONBLOBSTORE =
+ buildConf("spark.sql.hive.supportedSchemesToUseNonBlobstore")
+ .doc("Comma-separated list of supported blobstore schemes (e.g.
's3,s3a'). " +
+ "If any blobstore schemes are specified, this feature is enabled. " +
+ "When writing data out to a Hive table, " +
+ "Spark writes the data first into non blobstore storage, and then
moves it to blobstore. " +
+ "By default, this option is set to empty. It means this feature is
disabled.")
+ .version("3.1.0")
+ .stringConf
+ .createWithDefault("")
Review comment:
Users can specify any blob storage schema like following. If copy
operation is expensive in the storage system, this option will be effective.
- Amazon S3: `s3`, `s3a`, `s3n`
- Azure Blob Storage: `wasb`, `wasbs`
- Google Cloud Storage: `gs`
- Databricks: `dbfs`
- OpenStack: `swift`
Since any schemes are possible to be used, I believe we cannot define
specific supported schemes here. That's why I just listed samples in
SQLConf.scala.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
