[ https://issues.apache.org/jira/browse/TINKERPOP-2941?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17724398#comment-17724398 ]
ASF GitHub Bot commented on TINKERPOP-2941: ------------------------------------------- kenhuuu commented on code in PR #2053: URL: https://github.com/apache/tinkerpop/pull/2053#discussion_r1199338778 ########## hadoop-gremlin/src/main/java/org/apache/tinkerpop/gremlin/hadoop/Constants.java: ########## @@ -73,6 +73,7 @@ private Constants() { public static final String GREMLIN_SPARK_PERSIST_STORAGE_LEVEL = "gremlin.spark.persistStorageLevel"; public static final String GREMLIN_SPARK_SKIP_PARTITIONER = "gremlin.spark.skipPartitioner"; // don't partition the loadedGraphRDD public static final String GREMLIN_SPARK_SKIP_GRAPH_CACHE = "gremlin.spark.skipGraphCache"; // don't cache the loadedGraphRDD (ignores graphStorageLevel) + public static final String GREMLIN_SPARK_DONT_DELETE_NONE_EMPTY_OUTPUT = "gremlin.spark.dontDeleteNonEmptyOutput"; // don't delete the output if it is not empty Review Comment: Nit: I think there is a typo here. It should probably be GREMLIN_SPARK_DONT_DELETE_NON_EMPTY_OUTPUT > DO NOT purge the output location if it has content in SparkGraphComputer > ------------------------------------------------------------------------ > > Key: TINKERPOP-2941 > URL: https://issues.apache.org/jira/browse/TINKERPOP-2941 > Project: TinkerPop > Issue Type: Improvement > Affects Versions: 3.6.2 > Reporter: Redriver > Priority: Major > > The default logic for SparkGraphComputer is to purge all content if the > output location is specified. That is a dangerous operation especially for > the output location which contains contents. > https://github.com/apache/tinkerpop/blob/master/spark-gremlin/src/main/java/org/apache/tinkerpop/gremlin/spark/process/computer/SparkGraphComputer.java#L317:L324 > The correct behavior is to stop the process if it detects the output location > is not empty. -- This message was sent by Atlassian Jira (v8.20.10#820010)