On Mon, Aug 25, 2025 at 7:02 PM Mihail Nikalayeu <[email protected]> wrote: > > Amit Kapila <[email protected]>: > > > > What if the new insert happens in a page prior to the current page? I > > mean that the scan won't encounter the page where Insert happens. > > Hmm.... Yes - if the TID lands to the page left of the current > position, we’ll miss it as well. > A lock‑based solution (version in the v10) would require keeping all > pages with the same key under a read lock, which feels too expensive. >
Right. > > BTW, do we know the reason behind using SnapshotDirty in the first > > place? I don't see any comments in the nearby code unless I am missing > > something. > > I think this is simply an attempt to lock the newest version of the > logical tuple, including INSERT cases. > For an existing tuple, the same can be achieved using MVCC snapshot + retry. > However, in the case of a not-yet-committed INSERT, a different type > of snapshot is required. > > But I'm not sure if it provides any advantages. > I think it is better to document this race somewhere in a logical replication document for now unless we have a consensus on a way to move forward. -- With Regards, Amit Kapila.
