On Nov 20, 2007 9:18 AM, David Cantrell <david@cantrell.org.uk> wrote:
On Tue, Nov 20, 2007 at 06:49:53AM -0800, Anthony Ettinger wrote:

> /bin/cp -al /rem/.snapshots/hourly.0 /rem/.snapshots/hourly.1
> this is eating up 10% of MEM, and 2-3% of CPU, not too bad, but the process
> lasts over 5 minutes.

How long did it used to take, and what changed?



It's progressively gotten worse as the amount of data has grown, over the past year or two.

However, what I'm seeing now is that everything else on the system basically hangs until the copy operation is done. For example an unrelated database server and web server (locally) just freezes  when a cp request executes from the rsnapshot cron.

It has always had some slow down, but usually only a few seconds....30 at the most. Now it is over 4 minutes until I can type at the shell again.

--
Anthony Ettinger
Ph: 408-656-2473
var (bonita, farley) = new Dog;
farley.barks("very loud");
bonita.barks("at strangers");

http://chovy.dyndns.org/resume/
http://utuxia.com/consulting