Jan Wieck <JanWieck@Yahoo.com> writes:
> You're right ... forgot about that one.
> However, transactions from different origins are NEVER selected together
> and it wouldn't make sense to compare their xid's anyway. So the index
> might return index tuples for rows from another origin, but the
> following qualifications against the log_origin in the heap tuple will
> filter them out.
No, that's not the point here. The point here is that if the xids that
are simultaneously present in the index span more than a 2G-XID range,
btree *will fail* because it will be dealing with keys that do not obey
the transitive law. You do have a problem --- it doesn't explain Marc's
troubles, but sl_log_1_idx2 is broken for multi master situations. All
you need is masters with sufficiently different XID counters.
regards, tom lane
