[ https://issues.apache.org/jira/browse/PHOENIX-2716?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15170986#comment-15170986 ]
Gabriel Reid commented on PHOENIX-2716: --------------------------------------- [~jamestaylor] FWIW, I ran a test of this on a single-node cluster with an import of several million rows, with 4.5, 4.6, and the current master (i.e. just after the revert of PHOENIX-1973), and verified that performance of master is now pretty much in line with the performance in 4.5 (and that the work is appropriately being distributed over multiple reducers). I also verified that 4.6 was indeed broken (I believe [~sergey.soldatov] confirmed this as well on the mailing list). It would be great if someone could run this on a distributed env, but even failing that, I think this looks good for a new RC. > Performance regression for CSV bulk loader > ------------------------------------------ > > Key: PHOENIX-2716 > URL: https://issues.apache.org/jira/browse/PHOENIX-2716 > Project: Phoenix > Issue Type: Bug > Reporter: James Taylor > Assignee: Sergey Soldatov > Priority: Blocker > Fix For: 4.7.0 > > Attachments: PHOENIX-2716.patch > > > Looks like a serious performance regression in CSV bulk loading. See this > thread: https://s.apache.org/15Qb -- This message was sent by Atlassian JIRA (v6.3.4#6332)