Welcome to the sixteenth edition of the Kudu Weekly Update. This weekly blog post covers ongoing development and news in the Apache Kudu (incubating) project.
Development discussions and code in progress
-
Todd Lipcon changed the default bloom filter false positive (FP) ratio from 1% to 0.01%. The original value wasn’t scientifically chosen, but testing with billions of rows on a 5 node cluster showed a 2x insert throughput improvement at the cost of some more disk space.
-
J-D Cryans has been fixing some recently introduced bugs in the Java client. For example, see this patch and that one. Testability is a major concern right now in the Java client since triggering those issues requires a lot of time and data.
-
Dan Burkert has been making progress on the non-convering range partitioned tables front. The Java client now supports such tables and Dan is currently implementing new functionality to add and remove tablets via simple APIs.
-
David Alves is also making a lot of progress on the replay cache, this new component on the server-side which makes it possible for tablets to identify client write operations and enable exactly-once semantics. The main patch is up for review here.
-
Adar Dembo is working on addressing multi-master issues, as he explained in this blog post. He just put up for review patches that enable tablet servers to heartbeat to all masters. Part one is available here.
-
Misty prepared a document with J-D that contains instructions on how to release a new Kudu version. It is up for review here if you are curious or want to learn more about this process.
Project news
-
The vote to graduate Kudu from the ASF’s Incubator passed! The next step is for the ASF Board to vote on the resolution at their next meeting.
-
There’s a discussion on the dev mailing list about having an intermediate release, called 0.10.0, before 1.0.0. The current issue is that the version in the code is current “1.0.0-SNAPSHOT” which doesn’t leave room for another other release, but the bigger issue is that code is still churning a lot which doesn’t make for a stable 1.0 release.
Want to learn more about a specific topic from this blog post? Shoot an email to the kudu-user mailing list or tweet at @ApacheKudu. Similarly, if you’re aware of some Kudu news we missed, let us know so we can cover it in a future post.