Sign inSign up

loum/zeppelin-hive

By loum

Updated almost 5 years ago

Minimal Apache Zeppelin on Docker with Hive Interpreter capability.

Image
0

684

loum/zeppelin-hive repository overview

Apache Zeppelin 0.10.0|Apache Spark 3.1.2

Apache Zeppelin 0.9.0|Apache Hive 3.1.2

Quick reference

Quick start

Start the container
docker run --rm -d\
 --name zeppelin-hive\
 --hostname zeppelin-hive\
 --env ZEPPELIN_ADDR=0.0.0.0\
 --env ZEPPELIN_PORT=18888\
 --env ZEPPELIN_INTERPRETER_DEP_MVNREPO=https://repo1.maven.org/maven2/\
 --env ZEPPELIN_INTERPRETER_HIVE_DEFAULT_URL=jdbc:hive2://localhost:10000\
 --publish 18888:18888\
 loum/zeppelin-hive:latest

Zeppelin web UI is available at http://localhost:18888

In addition to the rich set of environment variables provided by Apache Zeppelin configuration you can also set the Hive interpreter's default.url via the custom ZEPPELIN_INTERPRETER_HIVE_DEFAULT_URL. For example, if your Hive server is running on the host named hive on port 10000 then set the ZEPPELIN_INTERPRETER_HIVE_DEFAULT_URL environment variable with this JDBC URL:

 ZEPPELIN_INTERPRETER_HIVE_DEFAULT_URL=jdbc:hive2://hive:10000
Terminate container
docker stop zeppelin-hive

Gotchyas

The custom hive interpreter overrides the zeppelin.jdbc.hive.timeout.threshold from 1 minute to 20 minutes. Not sure if it's related to https://issues.apache.org/jira/browse/ZEPPELIN-5146? Either way, this will hard-limit your query duration to twenty (20) minutes (as opposed to the default one (1) minute) regardless of the state of your query. If triggerred, the resultant "Cancel this job as no more log is produced in the last ?? seconds, maybe it is because no yarn resources" is an assumption that is just wrong. Worse, it manifests in your Spark logs as an obscure java.lang.NullPointerException: null error. I hope this is fixed or its intension made much, much clearer.

Tag summary

Content type

Image

Digest

Size

897.7 MB

Last updated

almost 5 years ago

docker pull loum/zeppelin-hive