MR3 1.7

Sungwoo Park

unread,

May 15, 2023, 11:18:45 AM5/15/23

to MR3

We are preparing the release of MR3 1.7 this week. For those using Hive-MR3-Kubernetes, I have uploaded a new Docker image mr3project/hive3:1.7, so you can try it now. (You don't have to change your current configurations.)

Cheers,

--- Sungwoo

Message has been deleted

Ill

unread,

May 18, 2023, 10:10:22 AM5/18/23

to MR3

Hi.

I noticed that MR3 1.7 has been released. But I didn't find its Release Notes. .

Can you add a ReleaseNotes to explain the new features added in version 1.7?

Also, what is the difference between "hivemr3-1.7-hive3.1.3-k8s.tar.gz" and "hivemr3-1.7-hive3.1.3-process.tar.gz"?

Ill

unread,

May 18, 2023, 10:24:41 AM5/18/23

to MR3

Hi. I got this error when I started 1.7 with the 1.6 configuration file.

Caused by: java.lang.IllegalArgumentException: For input string: "${tez.use.minimal}"

What do I need to adjust for version 1.6 of the configuration file?

Sungwoo Park

unread,

May 18, 2023, 10:46:33 AM5/18/23

to MR3

Hi,

Please remove from mr3-site.xml:

<property>
<name>mr3.cluster.use.hadoop-libs</name>
<value>${tez.use.minimal}</value>
</property>

TEZ_USE_MINIMAL in env.sh is no longer used.

There is not much in the release notes (except for the list of patches backported in MR3 1.7). Let me announce the release after updating all the documentation.

Cheers,

--- Sungwoo

Sungwoo Park

unread,

May 29, 2023, 8:57:01 AM5/29/23

to MR3

MR3 1.7 has been released. I have updated the binary releases at Github and the Docker image at Dockerhub.

For release notes, please see:

https://mr3docs.datamonad.com/docs/release/

https://mr3docs.datamonad.com/docs/release/#patches-backported-in-mr3-17

== Standalone mode ==

MR3 now supports standalone mode. You can run Hive on MR3 in a cluster of nodes, without having to install Hadoop or Kubernetes. For quick start guide, please see:

https://mr3docs.datamonad.com/docs/quick/standalone/

From TPC-DS 10TB benchmark, Hive 3 on MR3 runs faster than Hive-LLAP (8074s vs 8680s). Let me report the results in a blog article sometime soon.

From MR3 1.7, Hive on MR3 can run with Java 17. From the same TPC-DS benchmark, Java 17 reduces the total execution time by about 800s (from 8074s to 7415s).