Vincent Tran has posted comments on this change. ( http://gerrit.cloudera.org:8080/10550 )
Change subject: IMPALA-7078: Part 1: improve memory consumption of wide Avro scans ...................................................................... Patch Set 11: (2 comments) http://gerrit.cloudera.org:8080/#/c/10550/11/be/src/exec/hdfs-scan-node.cc File be/src/exec/hdfs-scan-node.cc: http://gerrit.cloudera.org:8080/#/c/10550/11/be/src/exec/hdfs-scan-node.cc@192 PS11, Line 192: nit: Intentional newline? http://gerrit.cloudera.org:8080/#/c/10550/11/be/src/exec/scan-node.h File be/src/exec/scan-node.h: http://gerrit.cloudera.org:8080/#/c/10550/11/be/src/exec/scan-node.h@191 PS11, Line 191: /// Cumulative number of scanner threads created during the scan. Some may be created : /// and then destroyed, so this can exceed the peak number of threads. This is a great observability addition to this counter. But to someone who is a neophyte like myself, it doesn't exactly spell out its usefulness. Can you expand a little bit on it? Specifically its implications on the performance of a scan node when compared to the peak number of threads? (i.e. what if it is an order of magnitude higher than the peak number / what about when it is close to the peak number?) -- To view, visit http://gerrit.cloudera.org:8080/10550 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Iebd2600b4784fd19696c9b92eefb7d7e9db0c80b Gerrit-Change-Number: 10550 Gerrit-PatchSet: 11 Gerrit-Owner: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Dan Hecht <dhe...@cloudera.com> Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Vincent Tran <vtt...@cloudera.com> Gerrit-Comment-Date: Wed, 06 Jun 2018 19:28:05 +0000 Gerrit-HasComments: Yes