On 23.05.25 11:43, Tomas Vondra wrote:
>> We don't currently have anything in the release notes that calls this
>> out as a potential upgrading issue, so I propose the attached patch.
>>
> Seems reasonable, although maybe it should say
>
> ... so if the old cluster does not have checksums enabled ...
>
> instead of
>
> ... so if the old cluster was initialized without checksums ...
>
> I mean, what matters is the current state, not how it was initialized.
committed with that change