[jira] [Updated] (HIVE-25335) Unreasonable setting reduce number, when join big size table(but small row count) and small size table
[ https://issues.apache.org/jira/browse/HIVE-25335?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] ASF GitHub Bot updated HIVE-25335: -- Labels: pull-request-available (was: ) > Unreasonable setting reduce number, when join big size table(but small row > count) and small size table > -- > > Key: HIVE-25335 > URL: https://issues.apache.org/jira/browse/HIVE-25335 > Project: Hive > Issue Type: Improvement >Reporter: zhengchenyu >Assignee: zhengchenyu >Priority: Major > Labels: pull-request-available > Attachments: HIVE-25335.001.patch > > Time Spent: 10m > Remaining Estimate: 0h > > I found an application which is slow in our cluster, because the proccess > bytes of one reduce is very huge, but only two reduce. > when I debug, I found the reason. Because in this sql, one big size table > (about 30G) with few row count(about 3.5M), another small size table (about > 100M) have more row count (about 3.6M). So JoinStatsRule.process only use > 100M to estimate reducer's number. But we need to process 30G byte in fact. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (HIVE-25335) Unreasonable setting reduce number, when join big size table(but small row count) and small size table
[ https://issues.apache.org/jira/browse/HIVE-25335?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] zhengchenyu updated HIVE-25335: --- Attachment: HIVE-25335.001.patch > Unreasonable setting reduce number, when join big size table(but small row > count) and small size table > -- > > Key: HIVE-25335 > URL: https://issues.apache.org/jira/browse/HIVE-25335 > Project: Hive > Issue Type: Improvement >Reporter: zhengchenyu >Assignee: zhengchenyu >Priority: Major > Attachments: HIVE-25335.001.patch > > > I found an application which is slow in our cluster, because the proccess > bytes of one reduce is very huge, but only two reduce. > when I debug, I found the reason. Because in this sql, one big size table > (about 30G) with few row count(about 3.5M), another small size table (about > 100M) have more row count (about 3.6M). So JoinStatsRule.process only use > 100M to estimate reducer's number. But we need to process 30G byte in fact. -- This message was sent by Atlassian Jira (v8.3.4#803005)