Details
Description
https://github.com/apache/spark/blob/master/graphx/src/main/scala/org/apache/spark/graphx/impl/EdgePartitionBuilder.scala#L48
in function toEdgePartition, code snippet " index.update(srcIds(0), 0)"
the elements in array srcIds are all not initialized and are all 0. The effect is that if vertex ids don't contain 0, indexSize is equal to (realIndexSize + 1) and 0 is added to the `index`
It seems that all versions have the problem
Attachments
Issue Links
- duplicates
-
SPARK-4173 EdgePartitionBuilder uses wrong value for first clustered index
- Resolved
- links to