Thread: [HACKERS] Partition : Append node over a single SeqScan
Hi All, Today while exploring a bit on Range table partitioning, I could see that even if scan is performed on a single partition, the plan node has Append node in it. Isn't it a bug? As per the usage of Append Node, it should only be seen in the queryplan when scan is performed on multiple tables. Following are the steps to reproduce the problem. CREATE TABLE part_tab (c1 int, c2 int) PARTITION BY RANGE (c1); CREATE TABLE part1 PARTITION OF part_tab FOR VALUES FROM (0) TO (100); CREATE TABLE part2 PARTITION OF part_tab FOR VALUES FROM (100) TO (200); postgres=# explain analyze select * from part_tab where c1 > 100; QUERY PLAN --------------------------------------------------------------------------------------------------------Append (cost=0.00..38.25rows=753 width=8) (actual time=0.009..0.009 rows=0 loops=1) -> Seq Scan on part2 (cost=0.00..38.25 rows=753 width=8) (actual time=0.009..0.009 rows=0 loops=1) Filter: (c1 > 100)Planning time: 166698.973 msExecution time: 0.043 ms (5 rows) -- With Regards, Ashutosh Sharma EnterpriseDB:http://www.enterprisedb.com
On Wed, Jul 5, 2017 at 3:48 PM, Ashutosh Sharma <ashu.coek88@gmail.com> wrote: > Hi All, > > Today while exploring a bit on Range table partitioning, I could see > that even if scan is performed on a single partition, the plan node > has Append node in it. Isn't it a bug? No. See following comment from create_append_plan() 1045 /* 1046 * XXX ideally, if there's just one child, we'd not bother to generate an 1047 * Append node but just return the single child. At the moment this does 1048 * not work because the varno of the child scan plan won't match the 1049 * parent-rel Vars it'll be asked to emit. 1050 */ -- Best Wishes, Ashutosh Bapat EnterpriseDB Corporation The Postgres Database Company
Hi, On Wed, Jul 5, 2017 at 3:58 PM, Ashutosh Bapat <ashutosh.bapat@enterprisedb.com> wrote: > On Wed, Jul 5, 2017 at 3:48 PM, Ashutosh Sharma <ashu.coek88@gmail.com> wrote: >> Hi All, >> >> Today while exploring a bit on Range table partitioning, I could see >> that even if scan is performed on a single partition, the plan node >> has Append node in it. Isn't it a bug? > > No. See following comment from create_append_plan() > 1045 /* > 1046 * XXX ideally, if there's just one child, we'd not bother to > generate an > 1047 * Append node but just return the single child. At the > moment this does > 1048 * not work because the varno of the child scan plan won't match the > 1049 * parent-rel Vars it'll be asked to emit. > 1050 */ > Thanks for the information. -- With Regards, Ashutosh Sharma EnterpriseDB:http://www.enterprisedb.com