Thread: BUG #8544: High nfs getattr

The following bug has been logged on the website:

Bug reference:      8544
Logged by:          bentzi
Email address:      bentzi.mor@hp.com
PostgreSQL version: 9.2.3
Operating system:   Redhat 6.3
Description:

While running postgres with NFS (like on netappp) we get between 5K - 30K
OPS while the read+write is only 10-20 OPS, means all recognize as other in
netapp.
While looking on postgres linux machine nfsstat show high getattr.


How to reduce it to get good performance and reduce load on the filer?


Bentzi

Hi,

> While running postgres with NFS (like on netappp) we get between 5K - 30K
> OPS while the read+write is only 10-20 OPS, means all recognize as other in
> netapp.
> While looking on postgres linux machine nfsstat show high getattr.

You should enable attribute caching on the partition hosting postgres'
datadir.

Greetings,

Andres Freund

--
 Andres Freund                       http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services