Hello,
I have noticed that in 9.0.2 EXPLAIN ANALYZE takes considerably longer time to return than the actual query. I never
noticedthis in 8.3.
Example:
In 9.0.2 i get
dynacom=# EXPLAIN ANALYZE SELECT count(ms.id) from marinerstates ms,mariner m where ms.endtime IS NOT NULL AND
ms.marinerid=m.id;
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------
Aggregate (cost=4716.48..4716.49 rows=1 width=4) (actual time=413.378..413.379 rows=1 loops=1)
-> Hash Join (cost=788.77..4582.04 rows=53775 width=4) (actual time=54.029..330.286 rows=53845 loops=1)
Hash Cond: (ms.marinerid = m.id)
-> Seq Scan on marinerstates ms (cost=0.00..2583.33 rows=53775 width=8) (actual time=0.005..99.092
rows=53845loops=1)
Filter: (endtime IS NOT NULL)
-> Hash (cost=583.90..583.90 rows=16390 width=4) (actual time=54.011..54.011 rows=16390 loops=1)
Buckets: 2048 Batches: 1 Memory Usage: 577kB
-> Seq Scan on mariner m (cost=0.00..583.90 rows=16390 width=4) (actual time=0.004..27.587 rows=16390
loops=1)
Total runtime: 413.495 ms
dynacom=# \timing
Timing is on.
dynacom=# SELECT count(ms.id) from marinerstates ms,mariner m where ms.endtime IS NOT NULL AND ms.marinerid=m.id;
count
-------
53845
(1 row)
Time: 39,038 ms
So the actual time (39 ms) differs considerably from the EXPLAIN ANALYZE Total runtime (413 ms) (10 times bigger)
In 8.3 i get
postgres@dynacom=# EXPLAIN ANALYZE SELECT count(ms.id) from marinerstates ms,mariner m where ms.endtime IS NOT NULL AND
ms.marinerid=m.id;
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------
Aggregate (cost=5094.13..5094.14 rows=1 width=4) (actual time=84.180..84.180 rows=1 loops=1)
-> Hash Join (cost=869.15..4954.27 rows=55942 width=4) (actual time=11.284..72.630 rows=54168 loops=1)
Hash Cond: (ms.marinerid = m.id)
-> Seq Scan on marinerstates ms (cost=0.00..2756.50 rows=55942 width=8) (actual time=0.013..25.696
rows=54168loops=1)
Filter: (endtime IS NOT NULL)
-> Hash (cost=642.40..642.40 rows=18140 width=4) (actual time=11.256..11.256 rows=16442 loops=1)
-> Seq Scan on mariner m (cost=0.00..642.40 rows=18140 width=4) (actual time=0.007..5.762 rows=16442
loops=1)
Total runtime: 84.346 ms
(8 rows)
postgres@dynacom=# SELECT count(ms.id) from marinerstates ms,mariner m where ms.endtime IS NOT NULL AND
ms.marinerid=m.id;
count
-------
54168
(1 row)
Time: 42.224 ms
In 8.3 the Total runtime is only double the value of the actual time.
Is there anything that can shed some light on this?
--
Achilleas Mantzios