Thread (22 messages) 22 messages, 9 authors, 2009-11-04

Re: unbelievably bad performance: 2.6.27.37 and raid6

From: Andrew Dunn <hidden>
Date: 2009-11-01 23:55:24

Thanks for the update Neil, good to have something to look forward to.

I am using Ubuntu 9.10, hopefully the new kernel will be incorporated
sometime in the near future. In the mean time I will back everything up
and create the ARRAY all over again.

Thomas Fjellstrom wrote:
On Sun November 1 2009, NeilBrown wrote:
  
quoted
On Mon, November 2, 2009 6:41 am, Thomas Fjellstrom wrote:
    
quoted
On Sun November 1 2009, Andrew Dunn wrote:
      
quoted
Are we to expect some resolution in newer kernels?
        
I assume all of the new per-bdi-writeback work going on in .33+ will
have a
large impact. At least I'm hoping.

      
quoted
I am going to rebuild my array (backup data and re-create) to modify
the chunk size this week. I hope to get a much higher performance when
increasing from 64k chunk size to 1024k.

Is there a way to modify chunk size in place or does the array need to
be re-created?
        
This I'm not sure about. I'd like to be able to reshape to a new chunk
size
for testing.
      
Reshaping to a new chunksize is possible with the latest mdadm and
 kernel, but I would recommend waiting for mdadm-3.1.1 and 2.6.32.
With the current code, a device failure during reshape followed by an
unclean shutdown while reshape is happening can lead to unrecoverable
data loss.  Even a clean shutdown before the shape finishes in that case
might be a problem.
    
That's good to know. Though I'm stuck with 2.6.26 till the performance 
regressions in the io and scheduling subsystems are solved.

  
quoted
NeilBrown

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

    

  
-- 
Andrew Dunn
http://agdunn.net
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help