Hi here, If I communicate quorums running zk 3.4 with a zk 3.5 client, when I am trying to create node in CONTAINER mode the quorums will compliant with KeeperException.UnimplementedException.
That is to say, as a client application, if I upgrade zk to 3.5 and then make use of this new feature, we doesn't have a fallback when zk server version is 3.4 to PERSISTENT or something like that, but just fail with an exception. Is it expected? If so, it seems upgrade client side force user to upgrade server side also. Best, tison. Zili Chen <wander4...@gmail.com> 于2019年8月26日周一 下午8:47写道： > Thanks for your insight Andor! > > I'll checkout the page :-) > > Best, > tison. > > > Andor Molnar <an...@apache.org> 于2019年8月26日周一 下午8:41写道： > >> Hi Zili, >> >> There’s no migration guide available for 3.5, because it shouldn’t break >> any existing functionality and no need to upgrade the database either. >> >> I’ve created a wiki page to collect upgrade experiences from users which >> could give you some hint if you’re facing problems: >> https://cwiki.apache.org/confluence/display/ZOOKEEPER/Upgrade+FAQ >> >> You can always drop an email here too to get help. >> >> Andor >> >> >> >> > On 2019. Aug 26., at 14:12, Zili Chen <wander4...@gmail.com> wrote: >> > >> > Detailedly, in Flink community we try to bump ZooKeeper version from >> 3.4.10 >> > to 3.5.5 but without accurate idea about how it would break existing >> > systems. >> > Mainly we make use of the "client" of ZooKeeper. >> > >> > >> > Zili Chen <wander4...@gmail.com> 于2019年8月26日周一 下午8:02写道： >> > >> >> Hi, >> >> >> >> Is there any migration guide for potentially breaking changes and how >> to >> >> deal with them? >> >> >> >> Best, >> >> tison. >> >> >> >>
