The actual data returned is a tiny fraction of the total table volume.
Is there a way to force an index scan on the partitions in a
guaranteed manner without resorting to re-writing queries with the
union all on partitions.
Thank you
Sriram
On Wed, Oct 14, 2009 at 10:15 PM, Tom Lane wrote:
> A
On Thu, Oct 15, 2009 at 2:51 PM, Anj Adu wrote:
> This appears to be a bug in the optimizer with resepct to planning
> queries involving child partitions. It is clear that "any" index is
> being ignored even if the selectivity is high. I had to re-write the
> same query by explicitly "union-all" '
Hi
This query is doing a sequential scan on the child partitions even
though indexes on all constrained columns are present
The box is very lightly loaded (8 core 15K 6x300G Raid 10 disks)
explain analyze
select thedate,sent.theboxid_id,sub_box_id,box_num,sum(summcount) as
event_count,'ACC'
This appears to be a bug in the optimizer with resepct to planning
queries involving child partitions. It is clear that "any" index is
being ignored even if the selectivity is high. I had to re-write the
same query by explicitly "union-all" ' ing the queries for individual
partitions.
On Wed, Oct
Anj Adu writes:
> This query is doing a sequential scan on the child partitions even
> though indexes on all constrained columns are present
It looks to me like it's doing exactly what it is supposed to, ie,
indexscan on the partitions where it would help and seqscans on the
partitions where it w
Hi
This query is doing a sequential scan on the child partitions even
though indexes on all constrained columns are present
The box is very lightly loaded (8 core 15K 6x300G Raid 10 disks)
explain analyze
select thedate,sent.theboxid_id,sub_box_id,box_num,sum(summcount) as
event_count,'ACC'
