Unfortunately the quick fix introduces a new config option which later needs to be deprecated if we implement the long term solution. If we ever implement the long term solution. That’s exactly what I wanted to discuss in the community since I cannot do anything with the current patch while a committer has -1 on it.
I don’t think that ZOOKEEPER-4874 should necessarily a big change, but let’s see other people’s opinion. Are u able to share numbers from your production environment? What’s the Jute buffer size between quorum members and how many ephemeral znodes cause the crash? Sharing more details about your usecase might also be beneficial. Why do your clients create so many ephemerals? Andor > On Aug 12, 2026, at 13:29, Dmitrii Duzhinskii <[email protected]> wrote: > > Hi Andor, > > The point I'd like to stress: this breaks production today, while the fix has > been stalled for years waiting for ZOOKEEPER-4874 — which may well be the > right long-term answer, but is a big change that won't be delivered any time > soon. > > Meanwhile the only real options I have are disabling closeSessionTxn (which > brings back ZOOKEEPER-3145) or raising jute.maxbuffer (which just moves the > threshold). Both are workarounds for a problem that ZOOKEEPER-3145 itself > introduced. > > Can't we treat #1716 as a fast fix for 3.9 and keep ZOOKEEPER-4874 as the > long term track? > > Dmitrii > >> I've refreshed my memory about the issue and the proposed patch. Looks like >> we didn’t reach a clear agreement on how to resolve the problem. The patch >> is about introducing a configurable limit to the number of ephemeral znodes, >> while @kezhu suggested [1] to get rid of fuzzy snapshots based on [2]. >> >> #1716 has a -1 from @eolivelli, so I cannot immediately merge it. >> >> Can we come back to this problem as a team and discuss how to move forward? >> Is there anybody who’s willing to work on ZOOKEEPER-4874? >> Do you think ZOOKEEPER-4874 would be acceptable to backport to branch-3.9? >> >> Thanks, >> Andor >> >> [1] https://github.com/apache/zookeeper/pull/1716#issuecomment-2421056078 >> [2] https://issues.apache.org/jira/browse/ZOOKEEPER-4874 >> >>> On Aug 12, 2026, at 08:12, Дмитрий Дужинский <[email protected]> wrote: >>> >>> Hi all, >>> >>> We've hit ZOOKEEPER-4306 (CloseSessionTxn exceeding jute.maxbuffer takes >>> down the whole ensemble) twice in production, in two independent >>> deployments — I've also left a comment on the JIRA. >>> >>> PR #1716 has been open since 2021 and was rebased in Oct 2024 with a >>> branch-3.9 backport (#2201), but both seem to have stalled since. >>> >>> Is anything still blocking them? If the 2021 review comments are the >>> remaining concern, we'd be glad to help address them >>> >>> https://issues.apache.org/jira/browse/ZOOKEEPER-4306 >>> https://github.com/apache/zookeeper/pull/1716 >>> https://github.com/apache/zookeeper/pull/2201
