[tech] morwong NFS speed problem

[Alastair Irvine]

On Tue, 12 July, 2005 at 03:34:53PM +0800, Adrian Chadd wrote:
[snip]
> 
> I'm talking about being objective. This isn't an attack against you.

Fair enough.

[snip timings]
> The trouble with this is:
> 
> * you haven't taken into account local buffering;

Yes I have.  It's a file I know hasn't been read on either machine since
the last reboot.

> * you didn't tell us which share this was;

It's in my home directory.  

> * you only did the test once

Further tests would have read from the local buffer cache.  You'll note
that I reversed the order of the tests for the 2MiB file in order to rule
out caching issues on martello.

morwong has had very low load (<0.05) all day AFAICT.

[snip timings]
> 
> Again, buffer cache, no share information, one test.
> We don't know how busy things are.
> 
> The other things to take into account:
> 
> * mussel is a 700ish MHz PIII
> * morwong is a less fast alpha (as leighton said, a piece of poop)
> * mussel has >512mb ram
> * morwong has what? 256mb?

None of these seem wildly different enough to account a two-order-of-
magnitude difference in the times spent for I/O.

This thread isn't Sysadmin 101; remember that it simply began with me
pointing out a problem.

> 
> Is mutt on morwong compiled 32 or 64 bit?

/usr/local/bin/mutt:    COFF format alpha dynamically linked, demand paged
executable or object module not stripped - version 3.13-14 

Can't tell.

-- 
... "Why is American beer served cold?  So you can tell it from urine."
                                       -- David Moulton
 _____________________________________________________________________ 
|                                                                     |
|  -=*Alastair Irvine*=-   <http://www.ucc.gu.uwa.edu.au/~alastair/>  |
|  C-monkey/wanderer/board&RPGer/net-nut  alastair at ucc.gu.uwa.edu.au  |
|_____________________________________________________________________|