On Sun, Feb 13, 2011 at 8:08 PM, Melvin Davidson <melvin6925 at yahoo.com>wrote:

> Aleksey,
>
> I can't really say how much disk space PostgreSQL 8.4 will use when
> processing the 45GB table. However, I did go back and look at your initial
> report and disk configuration.
>
> # df -h
> Filesystem            Size  Used Avail Use% Mounted on
> /dev/sda3             117G  110G  1.3G  99% /
> /dev/sda1              99M   24M   71M  26% /boot
> tmpfs                 3.9G     0  3.9G   0% /dev/shm
>
> You indicated your master has a 58GB database. Right at the start it means
> you will be using about half the available space. Add to that WAL files,
> system files, PostgreSQL binaries, etc, and it's easy to see that you will
> be using a large percentage of the only disk available.
>
> The best I can recommend would be to try increasing checkpoint_segments (
> try 20) temporarily. However, you might also look into just adding a larger
> disk (250GB?) or at least replacing the 117GB disk with a larger one.
>

Dear Melvin,

  Point taken.  Thanks for your time to look at my issue.

  This is how my master's disk free looks:

Filesystem            Size  Used Avail Use% Mounted on
/dev/sda3             124G   87G   32G  74% /

  So does the database get bigger as it is transfered by Slony? In other
words, does the replication have a disk utilization overhead?   If so, how
large should the slave's disk be compared to the master's disk?

  I had already increased checkpoint_segments from the default 3 to 24.  I
tried doubling it to 48.  Now apparently my replication is back up.
(Bucardo check_postgres script runs OK, and I am seeing updates come
across.)

  However my database size on the slave is 100 GB, and on the master is 58
GB.  What happened? How do I get it back down?

/var/lib/pgsql/data/base/25219965 on the Slave contains 48 files, each 1.1
GB in size,
and an 8.0KB PGVERSION file.  but "du -sh 25219965" says 101GB.

I'm definitely going to upgrade the disk subsystem on both servers as soon
as possible (I've already requested the funding).  Thanks for the hint re
checkpoint_segments.

I don't understand how I went so quickly from the COPY failing to
replication being up to date.

I will have to check tomorrow to make sure everything is there (on the
slave) that should be there.

Thanks!!  I really appreciate the help.
Aleksey
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20110213/abf552e5/attachment.htm