On Thu, 30 Jul 2009, Rauan Maemirov wrote:
maintenance_work_mem = 1GB work_mem = 192MB shared_buffers = 7680MB max_connections = 80 My box is Nehalem 2xQuad 2.8 with RAM 32Gb
While it looks like you sorted out your issue downthread, I wanted to point out that your setting for work_mem could be dangerously high here and contribute to problems with running out memory or using swap. If each of your 80 clients was doing a sort at the same time, you'd be using 80 * 192MB + 7680MB = 15360GB of RAM just for the server. The problem is that each client could do multiple sorts, so usage might even got higher. Unless you have a big data warehouse setup, more common work_mem settings are in the 16-64MB range rather than going this high. Just something to keep an eye on if you find a lot of memory is being used by the database processes. I really need to refine the pgtune model to more carefully account for this particular problem, it's a bit too aggressive here for people who aren't proactively watching the server's RAM after changing the settings.
-- * Greg Smith gsmith@xxxxxxxxxxxxx http://www.gregsmith.com Baltimore, MD -- Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
