Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-05 Thread Yin Huai
yeah... when hive.auto.convert.join.noconditionaltask=false; we cannot merge multiple Map-joins into a single Map phase... I think that trunk should work when hive.auto.convert.join.noconditionaltask=true. Thanks, Yin On Wed, Sep 4, 2013 at 4:03 PM, Nathanial Thelen wrote: > Doing: > > set hiv

Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-04 Thread Nathanial Thelen
Doing: > set hive.auto.convert.join.noconditionaltask=false; makes it work (though it does way more map reduce jobs than it should). When I get some time I will test against the latest trunk. Thanks, Nate On Sep 3, 2013, at 6:09 PM, Yin Huai wrote: > Based on the log, it may be also relate

Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-03 Thread Yin Huai
Based on the log, it may be also related to https://issues.apache.org/jira/browse/HIVE-4927. To make it work (in a not very optimized way), can you try "set hive.auto.convert.join.noconditionaltask=false;" ? If you still get the error, give "set hive.auto.convert.join=false;" a try (it will turn of

Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-03 Thread Ashutosh Chauhan
Not sure about EMR. Your best bet is to ask on EMR forums. Thanks, Ashutosh On Tue, Sep 3, 2013 at 2:18 PM, Nathanial Thelen wrote: > Is there a way to run a patch on EMR? > > Thanks, > Nate > > On Sep 3, 2013, at 2:14 PM, Ashutosh Chauhan wrote: > > Fix in very related area has been checked i

Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-03 Thread Nathanial Thelen
Is there a way to run a patch on EMR? Thanks, Nate On Sep 3, 2013, at 2:14 PM, Ashutosh Chauhan wrote: > Fix in very related area has been checked in trunk today : > https://issues.apache.org/jira/browse/HIVE-5129 Likely that will fix your > issue. > Can you try latest trunk? > > Ashutosh >

Re: Problems with 0.11, count(DISTINCT), and NPE

2013-09-03 Thread Ashutosh Chauhan
Fix in very related area has been checked in trunk today : https://issues.apache.org/jira/browse/HIVE-5129 Likely that will fix your issue. Can you try latest trunk? Ashutosh On Tue, Sep 3, 2013 at 2:03 PM, Nathanial Thelen wrote: > I am running Hive in EMR and since upgrading to 0.11 from 0.8.

Problems with 0.11, count(DISTINCT), and NPE

2013-09-03 Thread Nathanial Thelen
I am running Hive in EMR and since upgrading to 0.11 from 0.8.1.8 I have been getting NullPointerExceptions (NPE) for certain queries in our staging environment. Only difference between stage and production is the amount of traffic we get so the data set is much smaller. We are not using any c