Apache Phoenix

Monday January 22, 2018

Announcing CDH compatible Phoenix 4.13.2 released

The Apache Phoenix team is pleased to announce the immediate availability of a CDH compatible 4.13.2 release. Apache Phoenix enables SQL-based OLTP and operational analytics for Apache Hadoop using Apache HBase as its backing store and providing integration with other projects in the Apache ecosystem such as Spark, Hive, Pig, Flume, and MapReduce. The 4.x releases are compatible with HBase 0.98, 1.1, 1.2 and 1.3.

Highlights of the release include:

  • Compatibility with CDH 5.11.2 release
  • New parcels directory can be used directly as parcel repository from Cloudera Manager
  • Numerous bug fixes above and beyond the 4.13.0 release

Source and binary downloads are available here.

Comments:

That's great news - will there also be other CDH versions supported soon? (as it is a specific 5.11.2 parcel, but what about cdh 5.13, 5.14, etc?)

Posted by Alexander Meier on March 06, 2018 at 01:11 PM GMT #

Hi Alexander. The plan is expanding support to other CDH versions for next releases of Apache Phoenix. We are still on the process of deciding what is the best approach to ensure maximum parcel compatibility. We are thinking in several options: (A) Releasing one single parcel generated with wider compatibility ( cloudera-labs phoenix style ) . The issue here is that all transitive dependencies packaged in phoenix fatjars would be specific to a certain cdh version (i.e. cdh 5.11.2) but would be running against a different cdh version (maybe cdh5.14.0) . There is a small chance of incompatibility across versions ( even when all of them are HBase 1.2 based ) . Also we wouldn't be running our IT against all these cdh versions. (B) Release several parcels specific to different cdh versions. That is the safest option for better compatibility as we would be shipping the exact same libraries in the parcel as used in that version of cdh. Thanks!

Posted by Pedro Boado on March 09, 2018 at 12:15 AM GMT #

Hi I have notices that there seems to be an issue with the Parcel. To be specific with this artifact: <dependency> <groupId>org.apache.phoenix</groupId> <artifactId>phoenix-spark</artifactId> <version>4.13.2-cdh5.11.2</version> </dependency> The following will break: 1) spark shell with the phoenix-spark jar 2) import org.apache.phoenix.spark._ 3) new ToDataFrameFunctions(null) 4) this will result in a noSuchMethodException on the constructor for DataFrameFunctions This can also be observed in other constellations as when using saveToPhoenix functionality on DataFrames.

Posted by Dominic Egger on March 14, 2018 at 08:23 AM GMT #

Hi Dominic, I'm not able to replicate your issue. Would you mind filing a JIRA ticket and/or raise the question in our user email list? I've just opened a shell in a cloudera quickstart and your example just works.. [cloudera@quickstart workspace]$ spark-shell --master local[4] --conf "spark.executor.extraClassPath=/opt/cloudera/parcels/APACHE_PHOENIX-4.13.2-cdh5.11.2.p0.0/lib/phoenix/phoenix-4.13.2-cdh5.11.2-client.jar" --driver-class-path /opt/cloudera/parcels/APACHE_PHOENIX-4.13.2-cdh5.11.2.p0.0/lib/phoenix/phoenix-4.13.2-cdh5.11.2-client.jar

Posted by Pedro Boado on March 14, 2018 at 08:20 PM GMT #

Post a Comment:
  • HTML Syntax: NOT allowed

Calendar

Search

Hot Blogs (today's hits)

Tag Cloud

Categories

Feeds

Links

Navigation