[ https://issues.apache.org/jira/browse/TEZ-2000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14294875#comment-14294875 ]
Jeff Zhang commented on TEZ-2000: --------------------------------- [~sseth] hashCode() is not guaranteed to be unique. Here's one example I extracted from pig that 2 edges has different input vertex and output vertex, but has the same hashCode. {code} Vertex v1 = Vertex.create("scope-6", ProcessorDescriptor.create("p")); Vertex v2 = Vertex.create("scope-30", ProcessorDescriptor.create("p")); Vertex v3 = Vertex.create("scope-7", ProcessorDescriptor.create("p")); Vertex v4 = Vertex.create("scope-20", ProcessorDescriptor.create("p")); UnorderedKVEdgeConfig edgeConf1 = UnorderedKVEdgeConfig.newBuilder( Text.class.getName(), NullWritable.class.getName()).build(); Edge e1 = Edge .create(v1, v2, edgeConf1.createDefaultOneToOneEdgeProperty()); Edge e2 = Edge .create(v3, v4, edgeConf1.createDefaultOneToOneEdgeProperty()); LOG.info(e1.hashCode()); LOG.info(e2.hashCode()); {code} > Source vertex exists error during DAG submission > ------------------------------------------------ > > Key: TEZ-2000 > URL: https://issues.apache.org/jira/browse/TEZ-2000 > Project: Apache Tez > Issue Type: Bug > Affects Versions: 0.7.0 > Reporter: Rohini Palaniswamy > Assignee: Jeff Zhang > Attachments: TEZ-2000-1.patch > > > Pig e2e tests - Cross_5, Union_4 and Union_9 fail. This is due to some jira > that went in after TEZ-1931, but from the titles I cannot easily associate > one that could cause this failure. -- This message was sent by Atlassian JIRA (v6.3.4#6332)