[
https://issues.apache.org/jira/browse/HIVE-3632?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13714184#comment-13714184
]
Xuefu Zhang commented on HIVE-3632:
-----------------------------------
[~ashutoshc] My incomplete test shows that no upgrade script is necessary. The
change of column names was to fix mistakes in DN mapping file (package.jdo)
where a few columns weren't matched with the actually column names in schema.
Old DN didn't catch those, and worked magically. Thus, the patch is to fix the
field mapping rather than changing db schema.
One of my test was to create DB schema using hive-schema-0.10.0.mysql.sql, and
start metastore on the top of it. This patch works fine.
I said my test is incomplete as it didn't try in case of metastore schema auto
create. I don't think there will be a problem, but it's a good thing to check
too. Otherwise, I don't think there is a need for schema upgrade.
[~sergey.l.ivas...@intel.com] As to your question in RB, this patch doesn't
change connection pooling in any form. The change you observed is a "missing"
jar, which is part of DN-rdbms jar in the new DN version. Here is the DN doc
for your reference.
http://www.datanucleus.org/products/datanucleus/migration.html (search for
datanucleus-connectionpool).
I think Hive uses DBCP as a connection pool.
> Upgrade datanucleus to support JDK7
> -----------------------------------
>
> Key: HIVE-3632
> URL: https://issues.apache.org/jira/browse/HIVE-3632
> Project: Hive
> Issue Type: Bug
> Components: Metastore
> Affects Versions: 0.9.1, 0.10.0, 0.11.0
> Reporter: Chris Drome
> Assignee: Xuefu Zhang
> Priority: Critical
> Fix For: 0.12.0
>
> Attachments: HIVE-3632.patch
>
>
> I found serious problems with datanucleus code when using JDK7, resulting in
> some sort of exception being thrown when datanucleus code is entered.
> I tried source=1.7, target=1.7 with JDK7 as well as source=1.6, target=1.6
> with JDK7 and there was no visible difference in that the same unit tests
> failed.
> I tried upgrading datanucleus to 3.0.1, as per HIVE-2084.patch, which did not
> fix the failing tests.
> I tried upgrading datanucleus to 3.1-release, as per the advise of
> http://www.datanucleus.org/servlet/jira/browse/NUCENHANCER-86, which suggests
> using ASMv4 will allow datanucleus to work with JDK7. I was not successful
> with this either.
> I tried upgrading datanucleus to 3.1.2. I was not successful with this either.
> Regarding datanucleus support for JDK7+, there is the following JIRA
> http://www.datanucleus.org/servlet/jira/browse/NUCENHANCER-81
> which suggests that they don't plan to actively support JDK7+ bytecode any
> time soon.
> I also tested the following JVM parameters found on
> http://veerasundar.com/blog/2012/01/java-lang-verifyerror-expecting-a-stackmap-frame-at-branch-target-jdk-7/
> with no success either.
> This will become a more serious problem as people move to newer JVMs. If
> there are other who have solved this issue, please post how this was done.
> Otherwise, it is a topic that I would like to raise for discussion.
> Test Properties:
> CLEAR LIBRARY CACHE
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
