Gregory Stark <stark@enterprisedb.com> writes:
> Following up on some complaints we've had about extremely large plan times for
> large partitioned tables I've been doing some profiling.
> % cumulative self self total
> time seconds seconds calls s/call s/call name
> 62.58 203.76 203.76 21,474,937 0.00 0.00 SearchCatCache
> 9.01 233.08 29.32 1,923,789 0.00 0.00 list_nth_cell
> 5.34 250.46 17.38 2386465 0.00 0.00 pgstat_initstats
> 2.80 259.57 9.11 235691652 0.00 0.00 AllocSetAlloc
> 1.95 265.91 6.34 219852840 0.00 0.00 nocachegetattr
> 1.19 269.78 3.87 256569078 0.00 0.00 FunctionCall2
Is this supposed to be for a single run of the query? Cause the numbers
I get are a lot different:
% cumulative self self total time seconds seconds calls s/call s/call name
35.24 15.23 15.23 353301 0.00 0.00 SearchCatCache30.08 28.23 13.00 422469 0.00 0.00
list_nth_cell21.03 37.32 9.09 642488 0.00 0.00 pgstat_initstats 1.39 37.92 0.60 4281770
0.00 0.00 hash_search_with_hash_value 1.23 38.45 0.53 5130941 0.00 0.00 AllocSetAlloc 1.16
38.95 0.50 4154885 0.00 0.00 _bt_compare
(This is CVS HEAD as of today, which might or might not make a difference.)
regards, tom lane