Re: Physical sites handling large data

... that sound you hear is the sound of me knocking my head against the
brick wall in here...

Well it looks like Tom Lane was right (as always) on this one. On our
previous server, we had 4 Gigs of RAM and 1.6 Gigs of shared memory.
Does this mean now that the OS is efficiently caching disk, and they our
320MB of shared memory is good enough?

Our database is about 4 Gigs at this point with some tables having
hundreds of thousands or millions of records.

Running free looks like this.
[root@pg root]# free
             total       used       free     shared    buffers
cached
Mem:       5939524    5868720      70804          0      90732
5451808
-/+ buffers/cache:     326180    5613344
Swap:      2096440          0    2096440

There are 58 client processes running, with at times up to 220. The load
on this machine never runs more than 1 with Dual CPU's.

Top looks like this:
97 processes: 96 sleeping, 1 running, 0 zombie, 0 stopped
CPU0 states:  1.2% user,  3.2% system,  0.0% nice, 94.5% idle
CPU1 states:  0.1% user,  0.0% system,  0.0% nice, 99.4% idle
CPU2 states:  0.3% user,  0.2% system,  0.0% nice, 99.0% idle
CPU3 states:  0.3% user,  0.2% system,  0.0% nice, 99.0% idle
Mem:  5939524K av, 5874740K used,   64784K free,       0K shrd,   91344K
buff
Swap: 2096440K av,       0K used, 2096440K free                 5451892K
cached

Any definitive insight here as to why I'm running so well at this point?
- Ericson




On Mon, 2002-09-16 at 15:33, Manfred Koizar wrote:
> On 15 Sep 2002 11:33:59 -0400, Ericson Smith <eric@did-it.com> wrote:
> > shared memory to 3.2Gigs (out of 6GB Ram). [...]
> >shared_buffers = 38500
> >
> >ipcs output:
> >0x0052e2c1 98304      postgres  600        324018176  51
>
> Ericson, this looks more like 300MB to me; which might be a good
> choice anyway ;-)
>
> Servus
>  Manfred