[jira] [Commented] (DRILL-5214) This query cannot be planned possibly due to either a cartesian join or an inequality join
[ https://issues.apache.org/jira/browse/DRILL-5214?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16296371#comment-16296371 ] Dor commented on DRILL-5214: Hi, This is the query i did to get the above exception select cust.customer_id, cust.subscriber_id, resource_value from customers cust, USERS_ORACLE users where cust.key >= '0002:10:2017:0052' and cust.key < '0002:10:2017:0104' and cust.customer_id = users.customer_id and cust.subscriber_id = users.subscriber_id limit 100; Both users and customers tables are huge tables. > This query cannot be planned possibly due to either a cartesian join or an > inequality join > -- > > Key: DRILL-5214 > URL: https://issues.apache.org/jira/browse/DRILL-5214 > Project: Apache Drill > Issue Type: Bug >Reporter: Gautam Kumar Parai >Assignee: Gautam Kumar Parai > > We see the following error for certain queries: java.sql.SQLException: > UNSUPPORTED_OPERATION ERROR: This query cannot be planned possibly due to > either a cartesian join or an inequality join > This happens e.g. with TPCH Q11 with statistics [Drill-1328] -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (DRILL-6033) Using Drill Hive connection to query an Hbase table
Dor created DRILL-6033: -- Summary: Using Drill Hive connection to query an Hbase table Key: DRILL-6033 URL: https://issues.apache.org/jira/browse/DRILL-6033 Project: Apache Drill Issue Type: Bug Affects Versions: 1.11.0 Environment: 3 instances of Cloudera 5.10v , each one have a drillbit installed. Each machine has 24 vCPU. Reporter: Dor Using Drill hive connection to query Hbase table. +*Following query *+ select * from hive.mytable where key >= '0001:10:2017:0410:3157781' and key < '0001:10:2017:0410:3157782'; +*What happened*+ Failed with an error after timeout. It seems that the word 'key' didn't push down to hive from drill. +*What we also tried*+ Same query in Drill over hbase takes less than a sec, In hue hive it takes few seconds +*Debug trail*+ When you look in the sql profile of drill (using the web), you see a table full scan for millions of records, while actually it was supposed to return 9 rows. Does Drill on top of hive is using the key to access only the relevant region of the table? -- This message was sent by Atlassian JIRA (v6.4.14#64029)