Обсуждение: [HACKERS] Max value for -B

Поиск
Список
Период
Сортировка

[HACKERS] Max value for -B

От
The Hermit Hacker
Дата:
Hi...

    What is the max anyone has tried running postmaster with using the
- -B flag?  Under FreeBSD, I can't seem to get her above 256, and *believe*
I've up'd all the kernel config stuff, but before I actually debug that aspect,
figured I'd just check here and see if anyone else, on any of the other
platforms, has tried to go higher...and if it makes any difference?

Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org           secondary: scrappy@{freebsd|postgresql}.org

------------------------------

Re: [HACKERS] Max value for -B

От
"Micha³ Mosiewicz"
Дата:
The Hermit Hacker wrote:
>[...]
>         What is the max anyone has tried running postmaster with using the
> -B flag?  Under FreeBSD, I can't seem to get her above 256, and *believe*
> I've up'd all the kernel config stuff, but before I actually debug that aspect,
>[...]

I tried under Linux. 256 is a maximum. BTW. How does 'multiplexing
backends' work? Does it really work?

Mike

- --
********     MEMBER OF THE INTERNATIONAL PROGRAMMERS GUILD    ********
WWW: http://www.lodz.pdi.net/~mimo  tel: Int. Acc. Code + 48 42 148340
add: Michal Mosiewicz  *  Bugaj 66 m.54 *  95-200 Pabianice  *  POLAND

------------------------------

Re: [HACKERS] Max value for -B

От
"Thomas G. Lockhart"
Дата:
The Hermit Hacker wrote:
>         What is the max anyone has tried running postmaster with using the
> -B flag?

I just tried 300 and 512, with no apparent ill effects. Didn't do much
with the new parameters; createdb, destroydb, and the int8 user-package
and test inputs.

>  Under FreeBSD, I can't seem to get her above 256, and *believe*
> I've up'd all the kernel config stuff, but before I actually debug that aspect,
> figured I'd just check here and see if anyone else, on any of the other
> platforms, has tried to go higher...and if it makes any difference?

Don't know what "can't get above 256" actually means. Does postgres
refuse to start up, do you get core dumps, or are there more subtle
effects?? btw, I'm on a plain-vanilla dual Pentium Pro with 96MB of
(new) memory running Linux (don't choke... :)

            - Tom

------------------------------

Re: [HACKERS] Max value for -B

От
"Micha³ Mosiewicz"
Дата:
Micha� Mosiewicz wrote:
>[...]
> I tried under Linux. 256 is a maximum. BTW. How does 'multiplexing
> backends' work? Does it really work?

Oops. Correction - it works with more than 256. (Tried 300 and 512) But
I haven't tested it much... I remember that I had some problems with
older versions and -B > 256.

Mike

- --
********     MEMBER OF THE INTERNATIONAL PROGRAMMERS GUILD    ********
WWW: http://www.lodz.pdi.net/~mimo  tel: Int. Acc. Code + 48 42 148340
add: Michal Mosiewicz  *  Bugaj 66 m.54 *  95-200 Pabianice  *  POLAND

------------------------------

Re: [HACKERS] Max value for -B

От
The Hermit Hacker
Дата:
On Mon, 9 Jun 1997, Micha� Mosiewicz wrote:

> Micha� Mosiewicz wrote:
> >[...]
> > I tried under Linux. 256 is a maximum. BTW. How does 'multiplexing
> > backends' work? Does it really work?
>
> Oops. Correction - it works with more than 256. (Tried 300 and 512) But
> I haven't tested it much... I remember that I had some problems with
> older versions and -B > 256.

    okay, then its just me missing a config value in my kernel...will
look at it from that perspective then :)  Thanks...

Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org           secondary: scrappy@{freebsd|postgresql}.org

------------------------------

Re: [HACKERS] Max value for -B

От
The Hermit Hacker
Дата:
On Mon, 9 Jun 1997, Thomas G. Lockhart wrote:

> The Hermit Hacker wrote:
> >         What is the max anyone has tried running postmaster with using the
> > -B flag?
>
> I just tried 300 and 512, with no apparent ill effects. Didn't do much
> with the new parameters; createdb, destroydb, and the int8 user-package
> and test inputs.
>
> >  Under FreeBSD, I can't seem to get her above 256, and *believe*
> > I've up'd all the kernel config stuff, but before I actually debug that aspect,
> > figured I'd just check here and see if anyone else, on any of the other
> > platforms, has tried to go higher...and if it makes any difference?
>
> Don't know what "can't get above 256" actually means. Does postgres
> refuse to start up, do you get core dumps, or are there more subtle
> effects?? btw, I'm on a plain-vanilla dual Pentium Pro with 96MB of
> (new) memory running Linux (don't choke... :)

    Sorry Thomas...

    If I tried to run postmaster with a -B value of 512, the postmaster
wouldn't start up (but don't have an error message handy).  Since I've got
it confirmed that under Linux, at least, it will run iwth a value of 512,
then the problem is OS specific, not PostgreSQL, so will look at it from
that angle :)

    Now, with that in mind, what would be *really* neat (hey Webmaster,
getting this? *grin*) would be a chart of recommended values available from
our WWW site?

    For instance, if I have 48meg of RAM on my machine, what is a good
- -B value to use for optimal performance?  And, from that perspective, going
towards the idea of 'benchmarks'...what sort of test could we create that
would test that value?

Marc G. Fournier
Systems Administrator @ hub.org
primary: scrappy@hub.org           secondary: scrappy@{freebsd|postgresql}.org

------------------------------