Re: Performance degradation in 324577f39bc8738ed0ec24c36c5cb2c2f81ec660

Vladimir Kamarzin <vvk(at)vvk(dot)pp(dot)ru> writes:
> After upgrade from 9.3.1 to 9.3.5 we expirienced a slight performance degradation of all queries. Query time increased to some amount of ms, mostly in range of 100ms. Some actions in our application results in a lot of small queries and in such cases performance degradation is very significant - total action performs for a 2-3 times longer then before (15s -> 40s, etc).
> Using git-bisect we've found a bad revision causes performance drop: it is 324577f39bc8738ed0ec24c36c5cb2c2f81ec660

Hm. If you're going to do queries that involve update/delete across large
inheritance trees, that bug fix is unavoidably going to cost you some
cycles. Having said that, though, the append_rel_list data structures
aren't especially large or complex, so it's a mite astonishing that you
could notice this extra copying cost in the context of everything else
that happens in a large inherited UPDATE. I am wondering if you've
misidentified the commit that made the difference --- especially since you
claim there's a penalty for "all" queries, which there manifestly couldn't
be with this particular patch. If not, there must be something rather
unusual about your queries or schema. Care to provide a self-contained
test case?

regards, tom lane