hive-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Edward Capriolo <edlinuxg...@gmail.com>
Subject Re: Safe to upgrade to Hive 0.7.0?
Date Tue, 08 Feb 2011 00:30:25 GMT
On Mon, Feb 7, 2011 at 7:27 PM, Gokul Pillai <gokooltech@gmail.com> wrote:
> I have a Hadoop 0.20 cluster with Hive 0.6.
> I want to try out the ngrams functions in 0.7 (trunk) and so want to
> upgrade.
> What would you recommend. Can I have a Hive 0.7 install in a different
> location (for trial) and point it to the same cluster.
> If not, is there an upgrade utility?
> I am not concerned about existing tables. I just want to make sure I can run
> Hive 0.6 and Hive 0.7 side by side.
> Regards
> Gokul
>
> On Wed, Dec 8, 2010 at 9:26 AM, Edward Capriolo <edlinuxguru@gmail.com>
> wrote:
>>
>> On Wed, Dec 8, 2010 at 12:06 PM, Ning Zhang <nzhang@fb.com> wrote:
>> > These are datanucleus paramesters found in
>> > hive-default.xml/hive-site.xml. If you change them to the following the
>> > metastore schema won't be changed automatically (may be good for preventing
>> > accidents).
>> >
>> >
>> > <property>
>> >  <name>datanucleus.autoCreateSchema</name>
>> >  <value>false</value>
>> > </property>
>> >
>> > <property>
>> >  <name>datanucleus.fixedDatastore</name>
>> >  <value>true</value>
>> > </property>
>> >
>> >
>> > ________________________________________
>> > From: Edward Capriolo [edlinuxguru@gmail.com]
>> > Sent: Wednesday, December 08, 2010 7:19 AM
>> > To: user@hive.apache.org
>> > Cc: <hive-user@hadoop.apache.org>
>> > Subject: Re: Safe to upgrade to Hive 0.7.0?
>> >
>> > On Wed, Dec 8, 2010 at 9:55 AM, Ryan LeCompte <lecompte@gmail.com>
>> > wrote:
>> >> Hey guys,
>> >> We would like to experiment with some of the features in the current
>> >> trunk
>> >> (0.7) release. Do you think at this point it's nearing
>> >> "production-ready" ?
>> >> Are there any metastore migration scripts or anything like that before
>> >> upgrading to 0.7 from 0.6?
>> >> Thanks,
>> >> Ryan
>> >>
>> >
>> > Ryan,
>> >
>> > This is hard to say. There are many changes going in around indexing
>> > and authorization. IMHO everyone is taking a lot of care in the design
>> > however, it is still subject to change.
>> >
>> > If i were compelled to update I would spin up a new instance into a
>> > new meta-store back end. Then compare the schema of current with 0.6.X
>> > like a diff of 'show create tables'.
>> >
>> > Just an FYI. I applied two different hive trunks between 0.5 and 0.6.
>> > and I never needed the update scripts because hive happily updated the
>> > schema for me. After my final upgrade to 0.7.0 I re-ran everything in
>> > the upgrade script just to be safe. Nothing noticeably bad happened to
>> > me during the duration.
>> >
>> > I suspect many people will not need to run the update scripts since in
>> > the stock configuration hive seems to auto-update anyway. (Someone
>> > correct me if I am wrong here.) Does anyone know how to specifically
>> > which variable controls this?
>> >
>> > Edward
>> >
>>
>> https://issues.apache.org/jira/browse/HIVE-1841 -- please comment
>
>

I would not do it. The new version will attempt to update the
meta-store. I would just attempt to build n grams udf into a separate
JAR and load it up with create temporary function.

Mime
View raw message