It looks overkill to me to use a formal vote thread to start a release … and of course I support having this patch release soon.
I just skimmed the PR list and see unusual number of PR are labeled "correctness" while they should not, to clarify, correctness bug means it would silently corrupt the output data or produce wrong query result, for ordinal bugs which may fail the job or decrease performance, they should not be classified as correctness issue. Thanks, Cheng Pan > On Nov 9, 2026, at 17:20, Zhentao Shuai via dev <[email protected]> > wrote: > > Hi all, > > I'd like to start a new vote on resuming preparations for Apache Celeborn > 0.7.1 from branch-0.7. > > Since our earlier discussion, PR #3945 (CELEBORN-2539) has been merged into > main and backported to branch-0.7: > https://github.com/apache/celeborn/pull/3945 > > This fixes missing authorization for worker lifecycle and cluster management > operations on the Master's external RPC endpoint when authentication is > enabled. Authorization checks now cover both request/reply and one-way RPCs > before they can change cluster state. > > With this fix in place, I propose moving forward with the 0.7.1 maintenance > release, alongside the correctness and stability fixes already included in > branch-0.7. I volunteer to continue as release manager. > > Please vote on proceeding with release preparation: > [ ] +1 Proceed > [ ] 0 No opinion > [ ] -1 Do not proceed (please explain any remaining blockers) > > The vote will remain open for at least 72 hours. > > If we reach agreement, I'll coordinate any remaining backports and release > validation, prepare RC0, and start a separate vote on the release candidate. > > Thanks, > Zhentao Shuai
