On Wed, Dec 4, 2013 at 11:49 PM, Metin Doslu <metin@citusdata.com> wrote:
> Here are some extra information:
>
> - When we increased NUM_BUFFER_PARTITIONS to 1024, this problem is
> disappeared for 8 core machines and come back with 16 core machines on
> Amazon EC2. Would it be related with PostgreSQL locking mechanism?
I think here there is a good chance of improvement with the patch
suggested by Andres in this thread, but
still i think it might not completely resolve the current problem as
there will be overhead of associating data
with shared buffers.
Currently NUM_BUFFER_PARTITIONS is fixed, so may be auto tuning it
based on some parameter's can
help such situations.
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com