Re: Please help - tks

From: Brian McCane <bmccane(at)mccons(dot)net>
To: <pgsql-admin(at)postgresql(dot)org>
Subject: Re: Please help - tks
Date: 2001-10-19 14:42:52
Message-ID: 20011019093709.G71650-100000@fw.mccons.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

Brett,

It appears that I may have upset you somehow. I didn't intend to. I
chose to write the PERL script to place them in a transaction to help
improve performance a little bit on each pass. By building and tearing
down a connection to the postmaster everytime you send an update, you will
be putting a much heavier load on the server than is necessary. If I was
using the script myself, I would make the connection to the postmaster
once at the top of the script, and then re-use it with a transaction block
for each group of 10/20. Then shut it down at the end. This way I do not
suffer any additional protocol overhead each time I want to use it.

- brian

On Thu, 18 Oct 2001, Brett Schwarz wrote:

> ????
>
> This is why I said this:
> > > adjust the sleep to your own specs.
>
> If he wants to "batch" them, then all he needs to do is add a condition/count:
>
> --------------------------------
> #!/bin/sh
>
> cnt=1
> while read line; do
>
> # "batches" of 20 or whatever you *want*
> if [ "X$cnt" == "X20" ]; then
> sleep 5 # whatever you *want*
> cnt=0
> fi
>
> psql dbname -c "UPDATE ... $line ..."
> let cnt=$cnt+1
>
> done < filename
>
> ---------------------------------
>
> I will leave it up to the reader to modify this to his/her *specific* need...
>
> --brett
>
>
>
>
> On Thu, 18 Oct 2001 20:07:32 -0500 (CDT)
> Brian McCane <bmccane(at)mccons(dot)net> wrote:
>
> > Much too quick and dirty. With 300,000 updates this will take 1.5million
> > seconds, roughly 200 days. Even at a rate of 1 a second it will take
> > nearly four days. All this aside, I would probably use a perl script and
> > do batches of about 10-20, with a sleep. You can try something like the
> > following and feed it using standard input. Please NOTE, I didn't test
> > this, and even if I did, you shouldn't trust me :).
> >
> > - brian
> >
> > Wm. Brian McCane | Life is full of doors that won't open
> > Search http://recall.maxbaud.net/ | when you knock, equally spaced amid those
> > Usenet http://freenews.maxbaud.net/ | that open when you don't want them to.
> > Auction http://www.sellit-here.com/ | - Roger Zelazny "Blood of Amber"
> >
> > #! /usr/bin/perl
> >
> > use strict ;
> >
> > my @updates ;
> >
> > while (my $line = <STDIN>)
> > {
> > my $cmd = "UPDATE table SET ......." ; # Use the line we just read
> > push(@updates, $cmd) ;
> > if (@updates == 10)
> > {
> > do_update() ;
> > sleep 5 ;
> > @updates = () ; # Redundant
> > }
> > }
> > do_update if (@updates) ;
> >
> > sub do_update
> > {
> > open PSQL, "| psql dbname" ;
> > print PSQL "begin transaction ;\n" ;
> > while (my $cmd = shift @updates)
> > {
> > print PSQL $cmd . "\n" ;
> > }
> > print PSQL "commit ;\n" ;
> > close PSQL ;
> > }
> >
> > __END__
> >
> >
> > On Thu, 18 Oct 2001, Brett Schwarz wrote:
> >
> > > A quick and dirty solution would be something like (untested):
> > >
> > > while read line; do
> > > psql dbname -c "update table ... use 'line' as your id and data"
> > > sleep 5
> > > done < filename
> > >
> > > where "filename" is the file with the ids in it. "dbname" is the name of your DB (of course!)
> > >
> > > I am not sure what your file looks like, but you can do something like this. It could be easy, or harder, depending on what your file has.
> > >
> > > adjust the sleep to your own specs.
> > >
> > > --brett
> > >
> > >
> > > P.S.
> > >
> > > You could test this out by doing:
> > >
> > > while read line; do
> > > psql dbname -c "select * from table where id=$line"
> > > sleep 5
> > > done < filename
> > >
> > >
> > >
> > > On Thu, 18 Oct 2001 16:35:01 +1000 (EST)
> > > Mr OCP <mr_ocp(at)yahoo(dot)com> wrote:
> > >
> > > > Hi Folks
> > > >
> > > > We have about 300,000 records to update, I have
> > > > created a file with the ids of the records to be
> > > > updated, since the number of records to be updated is
> > > > very large, the update statement takes forever and
> > > > server slows down dramatically, therefore I want to do
> > > > one at a time.
> > > >
> > > > Do you guys have any script which may update records
> > > > one by one using the file that lists these records. I
> > > > am running postgresql sever under Unix, any kind of
> > > > script would do, whether its a perl, bash or sql
> > > > whatever script
> > > >
> > > > Thanks for your help
> > > > Mike
> > > >
> > > > http://briefcase.yahoo.com.au - Yahoo! Briefcase
> > > > - Manage your files online.
> > > >
> > > > ---------------------------(end of broadcast)---------------------------
> > > > TIP 5: Have you checked our extensive FAQ?
> > > >
> > > > http://www.postgresql.org/users-lounge/docs/faq.html
> > >
> > > _________________________________________________________
> > > Do You Yahoo!?
> > > Get your free @yahoo.com address at http://mail.yahoo.com
> > >
> > >
> > > ---------------------------(end of broadcast)---------------------------
> > > TIP 3: if posting/reading through Usenet, please send an appropriate
> > > subscribe-nomail command to majordomo(at)postgresql(dot)org so that your
> > > message can get through to the mailing list cleanly
> > >
> >
> >
> > ---------------------------(end of broadcast)---------------------------
> > TIP 6: Have you searched our list archives?
> >
> > http://archives.postgresql.org
>
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
>

In response to

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Allan Engelhardt 2001-10-19 15:02:38 [repost] pg_restore doesn't work with custom format?
Previous Message Eric Sawler 2001-10-19 13:10:35 RedHat logging