Re: [VOTE] Apache HAWQ 2.0.0.0-incubating RC3

2016-09-21 Thread Ed Espino
FYI: the two issues related to the Perl Artistic License have been resolved in both 2.0.0.0-incubating and master branches: Remove ASF Category X incompatible JSON Perl Module (Artistic license) https://issues.apache.org/jira/browse/HAWQ-1062 Declare PL/Perl's ppport.h source file to be an

Re: Re: External scan error: There are more external files (URLs) than primary segments that can read them (COptTasks.cpp:1756)

2016-09-21 Thread Vineet Goel
Your default_hash_table_bucket_number value is set to 6. Typically, this should be adjusted as 6 x #_of_your_segment_hosts. With 3 segments, you should set this value to 18. Any time you change this parameter, you should redistribute your HASH distributed tables, if you have any (unless the table

Re: [VOTE] Apache HAWQ 2.0.0.0-incubating RC3

2016-09-21 Thread Goden Yao
Just to confirm based on Legal-79, ppport.h is an exception with your PR, it looks we're good now. Thanks Ed. -Goden On Wed, Sep 21, 2016 at 1:52 AM Ed Espino wrote: > FYI: the two issues related to the Perl Artistic License have been resolved > in both 2.0.0.0-incubating and

Re: libhdfs3 development is still going on outside of ASF

2016-09-21 Thread Roman Shaposhnik
On Wed, Sep 21, 2016 at 10:08 AM, Kyle Dunn wrote: > What have we decided here? Fold libhdfs3 back into HAWQ for the near term > and revisit spinning it out in a dedicated submodule / repo down the road? To me this sounds like a near term decision. IOW, the other repo is

Re: Re: External scan error: There are more external files (URLs) than primary segments that can read them (COptTasks.cpp:1756)

2016-09-21 Thread Luis Macedo
Also on your location clause you should not reference the same file more than one time. If you want to scale gpfdist process you need to use different range on each port for a same server. (Not sure if I explain myself :)) If you use one gpfdist per server performance should be fine. One gpfdist

Re: libhdfs3 development is still going on outside of ASF

2016-09-21 Thread Kyle Dunn
What have we decided here? Fold libhdfs3 back into HAWQ for the near term and revisit spinning it out in a dedicated submodule / repo down the road? Do we need to have a consensus vote for this action? As for the outstanding PRs and issues in the current repo, who will be moving those to HAWQ?

[jira] [Created] (HAWQ-1066) Improper handling of install name for shared library on OS X

2016-09-21 Thread Kyle R Dunn (JIRA)
Kyle R Dunn created HAWQ-1066: - Summary: Improper handling of install name for shared library on OS X Key: HAWQ-1066 URL: https://issues.apache.org/jira/browse/HAWQ-1066 Project: Apache HAWQ

回复:Re: External scan error: There are more external files (URLs) than primary segments that can read them (COptTasks.cpp:1756)

2016-09-21 Thread 来熊
My environment is : 1 master 3 segments SQL >>: CREATE TABLE call_center ( cc_call_center_sk integer, cc_call_center_id character varying(16), cc_rec_start_date date, cc_rec_end_date date, cc_closed_date_sk integer, cc_open_date_sk integer, cc_name character

Re: New build time dependency: perl-JSON

2016-09-21 Thread Paul Guo
You are right, Ed, Thanks. 2016-09-21 12:34 GMT+08:00 Ed Espino : > Paul, > > That's what I thought until I performed a full build. You will notice the > following error when running calico.pl which is part of each build: > > make[4]: Entering directory >

Co-located Joins & Data Locality in HAWQ

2016-09-21 Thread Vineet Goel
Hi all, I have received a fair number of questions on the topic of handling data locality and co-located joins in HAWQ 2. Most of the questions are coming from the background where HAWQ 1.x defaulted to HASH distributed tables distributed by a key and hence resulted in local joins in most cases