Would anyone object to developing this in the trunk branch? Given that there's unlikely any minor releases before the completion of this project, keep the IPv6 development in a feature branch seems like overkill.
Let's shoot for Hadoop 3.6.0 release towards the end of the year? On Tue, Aug 18, 2026 at 2:26 PM Siyao Meng <[email protected]> wrote: > Hi Hadoop community, > > We would like to revive IPv6 support work in Hadoop. The immediate goal is > to address gaps in hadoop-common, hadoop-auth and hadoop-hdfs-client, > so that IPv6 support in some downstream projects (such as Ozone) can be > unblocked, which we estimated would take ~6 commits in hadoop. > > We plan to revisit the earlier work and turn the applicable parts into > small, > reviewable changes. The main areas would include address parsing and > formatting, networking, utilities, and test coverage for IPv4, dual-stack, > and IPv6-only environments. > > Feature branches HADOOP-11890 > <https://github.com/apache/hadoop/compare/HADOOP-11890> and HADOOP-17800 > <https://github.com/apache/hadoop/compare/HADOOP-17800> look too outdated > to work on or to be merged to trunk. Therefore, we will not work on either > feature > branch. But we might harvest useful bits from either and put that on trunk. > > After the relevant changes are merged and validated, downstream projects > will also need a Hadoop release that includes them. We would discuss and > plan for that release after the changes are merged. > > Feedback on the proposed scope and approach would be appreciated. Please > also let us know about any related work we should coordinate with, or if > you would like to contribute or review patches. > > ref: https://issues.apache.org/jira/browse/HADOOP-11890 > > Thanks, > Siyao > Hadoop committer, Ozone PMC member and committer > On behalf of Ozone PMC >
