Csaba Nagy wrote:
(before:)
> duration: 5437.499 ms statement: EXECUTE <unnamed> [PREPARE: SELECT
> ... ]
(after:)
> duration: 2200.570 ms statement: EXECUTE C_220047 [PREPARE: SELECT
> ... ]
Also.. as to why you see the behaviour change, I'd guess that you're
using log_min_duration and the plans for the queries where you are using
setFetchSize() have changed from one with low startup cost to one with
high startup cost due to switching from named to unnamed statements. The
high startup cost plan means the very first portal execution takes
longer and gets logged.. the low startup cost plan might be more
expensive overall but the cost is spread out across all the portal
executions, so no one execution takes long enough to get logged.
-O