Re: Problems with PG 9.3

[Dhruv Shukla <dhruvshukla82@xxxxxxxxx>]

Some more information about memory:

$free -m

             total       used       free     shared    buffers     cached

Mem:        387551     295326      92224          0         49     207903

-/+ buffers/cache:      87373     300177

Swap:         4095         46       4049

$ cat /proc/swaps

Filename                                Type            Size    Used    Priority

/dev/dm-1                               partition       4194296 48100   -1

DB resides on its own server and isnt hosted on a VM.

-Dhruv

On Mon, Aug 25, 2014 at 11:24 AM, Dhruv Shukla <dhruvshukla82@xxxxxxxxx> wrote:

Kevin,

Currently we have a max connection setting for 1000 connections.

And RAM on server is 384GB RAM.

4 Octa core CPU hyperthreaded so total 64 cores.

-Dhruv

On Mon, Aug 25, 2014 at 10:51 AM, Kevin Grittner <kgrittn@xxxxxxxxx> wrote:

Dhruv Shukla <dhruvshukla82@xxxxxxxxx> wrote:

> Other informational details about configurations are:

> shared_buffers = 80GB

> temp_buffers = 2GB
> work_mem = 2GB
> maintenance_work_mem = 16GB

Well, at the default max_connections of 100 that could easily
result in the server trying to allocate about 567GB of RAM.  If
your number of connections is more than 100, adjust proportionally
higher.  If the server doesn't have that, you could see extreme
swapping or various other problems as the OS tries to survive.

> effective_cache_size = 130GB

And the planner will be generating plans based on the assumption
that the combination of shared_buffers and the OS cache will have
this much RAM available for caching, so if you have max_connections
= 100 this configuration would only make sense on a machine with
RAM of 617GB or more, plus whatever is needed for the OS and
anything besides PostgreSQL that you want to run on the machine.

How much RAM is on the machine (or VM)?

--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

--
Regards
Dhruv
404-551-2578

--
Regards
Dhruv
404-551-2578