[postgis-users] Shared memory extesion on a a Mac for PostgreSQL/PostGIS

[William Kyngesburye]

On Mar 4, 2011, at 4:07 PM, Ture Pålsson wrote:

> 2011/3/4 Rosemary Alles <rosemary.alles at gmail.com>:
>> Hullo all,
>> 
>> I'm attempting to store several sets of GPS derived spatial data (simple
>> lat/lon mostly) in PostGIS and have hence installed PostgreSQL with the
>> PostGIS extension on my iMac (running 10.6.6), the readme that comes with
>> the PostgreSQL (at least for the latest version of it) casually warns of the
>> need to increase -significantly- the amounts of shared memory availalbe for
>> use. (See "attached" readme) Has anyone done this and what are the risks?
> 
> I played around with that a few months ago, and with my test case,
> pulling an Openstreetmap extract of Sweden into the database on a 2
> GByte iMac, I could not get any measurable performance improvement by
> increasing Postgres's shared-memory parameters. It seems that the
> OS-level buffer cache is doing a decent job. The obvious criticism
> against this test case is that it has a single database "user"
> executing a few large transactions which may not be a realistic
> scenario. Then again, I don't think there are any major risks changing
> the system coniguration as long as you stick to the recommended values
> and always think twice before hitting "return".

It's not really a "warning" about any potential problem.  It's an installation recommendation/requirement.  With the OS X defaults the Postgres server may not run at all, especially with large and/or many databases, and on 64bit processors.  It doesn't hurt the system.

Odd, I thought I had that info in my package readme.  I'll have to dig through my memory and notes to see why I removed it, or if it was accidental.

-----
William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
http://www.kyngchaos.com/

"Time is an illusion - lunchtime doubly so."

- Ford Prefect