hawq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ming LI (JIRA)" <j...@apache.org>
Subject [jira] [Created] (HAWQ-994) PL/R UDF need to be separated from postgres process for robustness
Date Wed, 10 Aug 2016 08:56:20 GMT
Ming LI created HAWQ-994:
----------------------------

             Summary: PL/R UDF need to be separated from postgres process for robustness
                 Key: HAWQ-994
                 URL: https://issues.apache.org/jira/browse/HAWQ-994
             Project: Apache HAWQ
          Issue Type: New Feature
            Reporter: Ming LI
            Assignee: Lei Chang


Background:
In previous single node DB, user always deploy testing code on another testing DB. Now the
data maintained in HAWQ grows enormously, so it is hard to deploy a testing hawq with the
same test data. 

So user need to run some testing UDF or deploy some UDFs which lack of testing the whole data
directly onto hawq in production env, which may crash in PL/R or R code. Sometimes poorly
written query leads to postmaster reset causing all running jobs to be cancelled and rolled
back. Customer often sees this as a HAWQ issue even if it is a user code issue. So we need
to separated from postgres process, and change inter process communication from shared memory
to others(e.g. pipe, socket and so on).




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message