I think after a discussion on #cassandra-dev yesterday, we are going to remove the requirement for schema agreement to deliver hints, as suggested by Jeff Jirsa.
Kind Regards, Brandon On Thu, Dec 19, 2024 at 7:43 AM Paul Chandler <p...@redshots.com> wrote: > > Hi Brandon, > > I am not sure which part changes after CASSANDRA-20118, there is still the > system mismatch going to CASSANDRA_4 caused by the change in > system.compaction_history, and going to UPGRADING, this is caused by the 2 > different sstable formats, so nothing that CASSANDRA-20118 fixes. > > So while CASSANDRA-20118 improves things, it does not fix these specific > issues, unless I have missed something? > > > On 19 Dec 2024, at 12:17, Brandon Williams <dri...@gmail.com> wrote: > > > > On Thu, Dec 19, 2024 at 4:11 AM Paul Chandler <p...@redshots.com> wrote: > >> C*4 -> CASSANDRA_4 : There is a schema mismatch, and hints are not sent > >> from C*4 node to C*5 nodes. > >> CASSANDRA_4 -> UPGRADING: Repairs are not possible and Nodes cannot be > >> added or replaced. > >> UPGRADING-> NONE: No issues. > > > > I'll note this will change after CASSANDRA-20118 > > > >> Any thoughts on whether having SCM controlled by JMX/nodetool is a good > >> idea? > > > > I think it's a good idea but it's tricky. As I said on 20118, "An > > unfortunate consequence of our use of static initialization is that > > once started, there is no way to change storage compatibility modes" > > and all the columns are defined statically, so that will have to be > > overcome. > > > > Kind Regards, > > Brandon >
