HDP 3.1 Released! All The Kafka!
HDP 3.1 Released! All The Kafka!
A major upgrade to Hadoop distribution has been released. Read on to learn how to upgrade to it.
Join the DZone community and get the full member experience.Join For Free
The upgrade documents to use can be found here.
I was able to quickly and easily upgrad from HDP 3.0.1 to HDP 3.1. This is the stable release you were looking for. This is the new Dockerized platform you need.
yum upgrade ambari-agent -y yum upgrade ambari-server -y
New HDP Features
HDP 3.1, Ambari 2.7.3, and SmartSense 1.5.1
Kafka Streams Now Supported
Kafka Streams is now officially supported. Kafka Streams is fully integrated with platform services like Ranger and Schema Registry and supports secure/unsecure clusters.
- Simplified access control management with support for prefixed ACLs for bulk management of entities. Access control for topic creation for specific topics/topic prefix.
- Hostname verification to prevent SSL configuration man in the middle attacks.
- Improved Encryption support with faster TLS and CRC32C implementations. Over-the-wire encryption is faster.
- Simpler security configuration with SSL trust stores update without broker restart and security for Zookeeper listeners can be configured before starting brokers.
- Quota limit notifications to distinguish network errors from quota limits reached.
- Better broker resiliency by reducing memory footprint of message down conversion.
- Replication protocol improvements for fixing log divergence during fast leader failover.
- Prevent indefinite consumer block with new configuration options.
- Windowed aggregation performance in Kafka streams vastly improved.
- Support for icons in lineage graph by type of entity (for example, Apache Hive tables, Apache Kafka topics, etc.).
- Filtering of lineage UI to exclude deleted entities and excluding process entities as an option.
- Support for expanding lineage graph in UI to different number of hops (3, 6, 9, 12, etc.).
- Various performance and stabilization fixes.
- Support Apache Kafka 2.0 in Kafka Ranger Plugin.
- Support 'DelegationToken' resource type with 'Describ' operation for Kafka.
- Support for cluster resource type for Kafka.
- Support for Create operation with Topic resource in Kafka.
- Read and write Apache Kafka topics via Kafka Storage Handler.
- Accelerating joins between Druid table and Hive table by runtime filtering of Druid tables using bloom filters constructed on Hive tables.
- Exploiting constraints to generate efficient query plan.
- Improvements in stat system to generate better query plans.
- Improvements in Map join vectorization and filter expressions.
- JDBCStorageHandler improvements for MySQL and Postgres.
Ambari 2.7.3 and SmartSense 1.5.1
The latest maintenance release of our open source management tool for provisioning, managing, and monitoring HDP contains a number of fixes. For more details on the specific fixes, please see the Ambari Release Notes. We encourage current users of Ambari 2.7 to upgrade to 2.7.3 to take advantage of these fixes.
SmartSense 1.5.1 is the latest maintenance release focused on bug fixes and diagnostics capture improvements. For more information about the fixes introduced with this release, please see the SmartSense Release Notes.
Open JDK8 Support
Official Apache component versions for HDP 3.1.0:
- Apache Accumulo 1.7.0
- Apache Atlas 1.1.0
- Apache Calcite 1.16.0
- Apache DataFu 1.3.0
- Apache Hadoop 3.1.1
- Apache HBase 2.0.2
- Apache Hive 3.1.0
- Apache Kafka 2.0.0
- Apache Knox 1.0.0
- Apache Livy 0.5.0
- Apache Oozie 4.3.1
- Apache Phoenix 5.0.0
- Apache Pig 0.16.0
- Apache Ranger 1.2.0
- Apache Spark 2.3.2
- Apache Sqoop 1.4.7
- Apache Storm 1.2.1
- Apache TEZ 0.9.1
- Apache Zeppelin 0.8.0
- Apache ZooKeeper 3.4.6
- Apache Superset 0.23.3
- Apache Druid 0.12.1
From improved UIs to faster performance to Dockerization to more features, you hav to try HDP 3.1 now.
Using Cloudbreak you can spin one up in your cloud of choice in minutes as a full web click experience. Start ingesting, running, querying and apply machine learning to data in real-time at massive scale. All of this is open source and available for you to use now.
Opinions expressed by DZone contributors are their own.