David Hinkle <hinkle@cipafilter.com> writes:
> Tom, there are three columns in this table that exhibit the problem,
> here is the statistics data after an analyze, and the real data to
> compare it to.
> attname | n_distinct | most_common_freqs
> titleid | 292 | {0.767167}
Ouch. That's saying there's some single value of titleid that accounts
for more than three-quarters of the entries ... does that square with
reality? That'd certainly explain why a hash join goes nuts.
regards, tom lane
