Jeff Frost <jeff(at)frostconsultingllc(dot)com> writes:
> Tom Lane wrote:
>> Okay. What that means is that the indexes were created on data that had
>> already been inserted and updated to some extent, resulting in
>> HOT-update chains that turned out to be illegal for the new indexes.
>> The way we deal with this is to mark the indexes as not usable by any
>> query that can still see the dead HOT-updated tuples.
> Any idea why I don't see it on 8.3.4?
I think it's more likely some small difference in your test conditions
than any real version-to-version difference. In particular I think the
"still see" test might be influenced by the ages of transactions running
concurrently.
regards, tom lane