Thread: too many users for postgresql? :)

too many users for postgresql? :)

From
will trillich
Date:
so there i was, minding my own business, browsing around, and at
the postgresql.org site today i got--

    Warning: pg_connect() unable to connect to PostgreSQL
    server: FATAL 1: Sorry, too many clients already in
    /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
    Unable to access database

1) that speaks well for the future of postgresql because there
are enough people interested and browsing that they overran the
current parameters. probably not hard to fix the postgresql end
on this.

2) it gives out lots more information than necessary -- now i
know where the docs are... maybe some php parameters need some
tweaking, eh?

:)

--
There are 10 kinds of people:
ones that get binary, and ones that don't.

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

Looking for a firewall? Do you think smoothwall sucks? You're
probably right... Try the folks at http://clarkconnect.org/ !

Re: too many users for postgresql? :)

From
Justin Clift
Date:
will trillich wrote:
> so there i was, minding my own business, browsing around, and at
> the postgresql.org site today i got--
>
>     Warning: pg_connect() unable to connect to PostgreSQL
>     server: FATAL 1: Sorry, too many clients already in
>     /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
>     Unable to access database

Hi Will,

Which page(s) were you at to get this?  Have been trying to temporarily
disable database access from parts of the site to stop this error until
the max_connections parameter is bumped up.  Looked like all the pages
haven't been fixed yet.

:-(

Regards and best wishes,

Justin Clift


> 1) that speaks well for the future of postgresql because there
> are enough people interested and browsing that they overran the
> current parameters. probably not hard to fix the postgresql end
> on this.
>
> 2) it gives out lots more information than necessary -- now i
> know where the docs are... maybe some php parameters need some
> tweaking, eh?
>
> :)
>


--
"My grandfather once told me that there are two kinds of people: those
who work and those who take the credit. He told me to try to be in the
first group; there was less competition there."
- Indira Gandhi


Re: too many users for postgresql? :)

From
will trillich
Date:
On Mon, Jan 27, 2003 at 04:20:52PM +1030, Justin Clift wrote:
> will trillich wrote:
> >so there i was, minding my own business, browsing around, and at
> >the postgresql.org site today i got--
> >
> >    Warning: pg_connect() unable to connect to PostgreSQL
> >    server: FATAL 1: Sorry, too many clients already in
> >    /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
> >    Unable to access database
>
> Hi Will,
>
> Which page(s) were you at to get this?  Have been trying to temporarily
> disable database access from parts of the site to stop this error until
> the max_connections parameter is bumped up.  Looked like all the pages
> haven't been fixed yet.

the specific page i was after was

    http://www.postgresql.org/idocs/index.php?xfunc-plhandler.html

so apparently <docroot>/index.php called (either directly or
through other modules) opendb.php which is where the trouble
arose.

no big deal for me -- i had more serious issues trying to get my
own server re-started after adding another hard drive with bad
parameters (which locked everything up so bad i quite nearly had
a nervous breakdown. thank goodness i thought to try
clarkconnect's rescue_ide boot on their install cd. everything
came back up after a few fsck's and a revisit to the autodetect
in bios... whew! very nearly had to call the men in white coats!)

--
There are 10 kinds of people:
ones that get binary, and ones that don't.

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

Looking for a firewall? Do you think smoothwall sucks? You're
probably right... Try the folks at http://clarkconnect.org/ !

Re: too many users for postgresql? :)

From
will trillich
Date:
> will trillich wrote:
> >so there i was, minding my own business, browsing around, and at
> >the postgresql.org site today i got--
> >
> >    Warning: pg_connect() unable to connect to PostgreSQL
> >    server: FATAL 1: Sorry, too many clients already in
> >    /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
> >    Unable to access database

and just now (Mon Jan 27 01:05:28 CST 2003) at the root level of
the postgresql.org website, i see lots of

    "temporarily unavailable"

under 'news' and 'events'... maybe i just happened to hit the
vacuum or dump cron job?

--
There are 10 kinds of people:
ones that get binary, and ones that don't.

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

Looking for a firewall? Do you think smoothwall sucks? You're
probably right... Try the folks at http://clarkconnect.org/ !

Re: too many users for postgresql? :)

From
Lamar Owen
Date:
On Monday 27 January 2003 00:50, Justin Clift wrote:
> will trillich wrote:
> > so there i was, minding my own business, browsing around, and at
> > the postgresql.org site today i got--

> >     Warning: pg_connect() unable to connect to PostgreSQL
> >     server: FATAL 1: Sorry, too many clients already in
> >     /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
> >     Unable to access database

> Which page(s) were you at to get this?  Have been trying to temporarily
> disable database access from parts of the site to stop this error until
> the max_connections parameter is bumped up.  Looked like all the pages
> haven't been fixed yet.

This is one of the biggest advantages to the pooled model of AOLserver.  If
you have the number of connections to the database set equal to the number of
connection threads, you can never get this sort of message.  The webserver
will refuse the connection first -- which is probably a better thing to do,
rather than give such a damaging ('What? PostgreSQL can't scale to the load
of its own server?') error message.  While you certainly can do this with
Apache/PHP, it requires a much larger number to make it work well under load.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11


Re: too many users for postgresql? :)

From
Lincoln Yeoh
Date:
At 11:46 PM 1/26/03 -0600, will trillich wrote:

>so there i was, minding my own business, browsing around, and at
>the postgresql.org site today i got--
>
>         Warning: pg_connect() unable to connect to PostgreSQL
>         server: FATAL 1: Sorry, too many clients already in
>         /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
>         Unable to access database
>
>1) that speaks well for the future of postgresql because there
>are enough people interested and browsing that they overran the
>current parameters. probably not hard to fix the postgresql end
>on this.

As there's no robots.txt file, ( http://www.postgresql.org/robots.txt =>
404) it could just be a spider/crawler (e.g. google) hitting the site with
many connections at once.

If that is an issue, I would suggest letting google et all, spider the
static docs, and disallowing them from the dynamic idocs. However this
would mean useful suggestions in the idocs won't turn up in google etc.

Link.


Re: too many users for postgresql? :)

From
will trillich
Date:
On Mon, Jan 27, 2003 at 08:57:39AM -0500, Lamar Owen wrote:
> > will trillich wrote:
> > >     Warning: pg_connect() unable to connect to PostgreSQL
> > >     server: FATAL 1: Sorry, too many clients already in
> > >     /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
> > >     Unable to access database

> This is one of the biggest advantages to the pooled model of
> AOLserver.  If you have the number of connections to the
> database set equal to the number of connection threads, you
> can never get this sort of message.  The webserver will refuse
> the connection first -- which is probably a better thing to
> do, rather than give such a damaging ('What? PostgreSQL can't
> scale to the load of its own server?') error message.

a MUCH better thing. :)

--
There are 10 kinds of people:
ones that get binary, and ones that don't.

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

Looking for a firewall? Do you think smoothwall sucks? You're
probably right... Try the folks at http://clarkconnect.org/ !

Re: too many users for postgresql? :)

From
will trillich
Date:
On Mon, Jan 27, 2003 at 10:03:28PM +0800, Lincoln Yeoh wrote:
> At 11:46 PM 1/26/03 -0600, will trillich wrote:
> >        Warning: pg_connect() unable to connect to PostgreSQL
> >        server: FATAL 1: Sorry, too many clients already in
> >        /usr/local/www/www.postgresql.org/idocs/opendb.php on line 3
> >        Unable to access database

> As there's no robots.txt file, (
> http://www.postgresql.org/robots.txt => 404) it could just be
> a spider/crawler (e.g. google) hitting the site with many
> connections at once.

good point. the webmaster(s) can check their logs to see what it
was that spiked it yesterday.

> If that is an issue, I would suggest letting google et all,
> spider the static docs, and disallowing them from the dynamic
> idocs. However this would mean useful suggestions in the idocs
> won't turn up in google etc.

maybe have a different "driver" return idocs for known spiders
(base content-spewed on user-agent); there's gotta be a way to
have both search-engine-able idocs and enough connections to
keep it open for the human browsers...

--
There are 10 kinds of people:
ones that get binary, and ones that don't.

will@serensoft.com
http://sourceforge.net/projects/newbiedoc -- we need your brain!
http://www.dontUthink.com/ -- your brain needs us!

Looking for a firewall? Do you think smoothwall sucks? You're
probably right... Try the folks at http://clarkconnect.org/ !

Re: too many users for postgresql? :)

From
Bruno Wolff III
Date:
On Mon, Jan 27, 2003 at 10:53:15 -0600,
  will trillich <will@serensoft.com> wrote:
>
> maybe have a different "driver" return idocs for known spiders
> (base content-spewed on user-agent); there's gotta be a way to
> have both search-engine-able idocs and enough connections to
> keep it open for the human browsers...

If the robots are well behaved you can control things through robots.txt.
If they are evil robots, you need to have something that limits request
rates for a single IP address (or maybe a /24).