On Mon, 2005-10-03 at 15:33 -0700, Doug Cutting wrote:
> Rod Taylor wrote:
> > Virtually no IO reported at all.  Averages about 200kB/sec read and
> > writes are usually 0, but burst to 120MB/sec for under 1 second once
> > every 30 seconds or so.
> 
> That's strange.  I wonder what it's doing.  Can you use 'kill -QUIT' to 
> get a thread dump?  Try a few of these to sample the stack and see where 
> it seems to be spending time.

The stacktrace is running FileOutputStream.writeBytes in the
LocalFileSystem.write function, so it is definitely IO bound.

I had the admins change a number of settings on a couple of machines
(ext3 fs to journal=writeback, noatime, etc.) and those machines have
completed their work in a much more reasonable timeframe. I guess we can
chalk it up to a lot of head movement

I guess bumping access times and things for all of the .out files had a
large impact.


Thanks for your patience. Is there a "best practises" page in the Wiki
that I could contribute this insight to?

-- 
Rod Taylor <[EMAIL PROTECTED]>

Reply via email to