hi all,
I suggest to simplify deployment of CarbonData as following. 1. remove kettle dependency completely, no need to deploy "carbonplugins" folder on each node, no need to set "carbhon.kettle.home" 2. remove carbon.properties file from executor side, pass CarbonData configuration to executor side from driver side 3. use "spark.sql.warehouse.dir"(spark2) or "hive.metastore.warehouse.dir"(spark1) instead of "carbon.storelocation" So we will just need to deploy CarbonData jars on cluster mode in the future. What's your opinion? Best Regards David Cai
Best Regards
David Cai |
Administrator
|
Hi
Thanks you started a good discussion. For 1 and 2, i agree. In 1.0.0 version, will support it. For 3 : Need keep the parameter, users can specify carbon's store location. If users don't specify the carbon store location, can use the default location what you suggested: "spark.sql.warehouse.dir"(spark2) or "hive.metastore.warehouse.dir"(spark1) Regards Liang
|
I suggest we have a parallel implementation by removing kettle and when it
stabilises we make kettle deprecated and then remove it On Mon, 26 Dec 2016 at 1:25 PM, Liang Chen <[hidden email]> wrote: > Hi > > Thanks you started a good discussion. > > For 1 and 2, i agree. In 1.0.0 version, will support it. > For 3 : Need keep the parameter, users can specify carbon's store location. > If users don't specify the carbon store location, can use the default > location what you suggested: "spark.sql.warehouse.dir"(spark2) or > "hive.metastore.warehouse.dir"(spark1) > > Regards > Liang > > QiangCai wrote > > hi all, > > > > I suggest to simplify deployment of CarbonData as following. > > 1. remove kettle dependency completely, no need to deploy > > "carbonplugins" folder on each node, no need to set "carbhon.kettle.home" > > 2. remove carbon.properties file from executor side, pass CarbonData > > configuration to executor side from driver side > > 3. use "spark.sql.warehouse.dir"(spark2) or > > "hive.metastore.warehouse.dir"(spark1) instead of "carbon.storelocation" > > > > So we will just need to deploy CarbonData jars on cluster mode in the > > future. > > > > What's your opinion? > > > > Best Regards > > David Cai > > > > > > -- > View this message in context: > http://apache-carbondata-mailing-list-archive.1130556.n5.nabble.com/Discussion-Simplify-the-deployment-of-carbondata-tp5000p5006.html > Sent from the Apache CarbonData Mailing List archive mailing list archive > at Nabble.com. > |
Free forum by Nabble | Edit this page |