opinion on XFS

Roger L. Smith roger at ERC.MsState.Edu
Fri May 10 06:06:41 PDT 2002


On Fri, 10 May 2002, Greg Lindahl wrote:

> On Thu, May 09, 2002 at 08:49:01PM -0500, Roger L. Smith wrote:
>
> > This is a bit of a circular argument.  "XFS doesn't solve problems, it
> > just runs better on hardware that ext2 didn't work correctly on."
>
> I didn't say the first half of the quote. It's more fair to say that I
> don't think that ext2 is broken in general. I only have suspicions why
> XFS worked for you when you think ext2 did not.

I wasn't intending the quotations to be a direct quotation of anyone, I
was merely paraphrasing what the argument that several people are making
sounds like from where I'm sitting.

> > Just so I don't sound like a complete XFS zealot (which I'm not), I do
> > have to say that I've been quite satisfied with the performance of ext3.
>
> ext3 should have all of the correctness problems you think you saw
> with ext2.

Possibly true.  I guess I should have pointed out that I have not tested
ext3 in the same problem scenario under which ext2 failed.  I eventually
reconfigured things so that all of our user data is now mounted from a
Solaris system (where I already had several terabytes of RAID storage
available).  The partition that I have shared among all of the nodes that
is ext3 now is mainly read-only, so it would be less prone to the issues
that I had seen before when it was used otherwise.

 _\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_\|/_
| Roger L. Smith                        Phone: 662-325-3625               |
| Systems Administrator                 FAX:   662-325-7692               |
| roger at ERC.MsState.Edu                 http://WWW.ERC.MsState.Edu/~roger |
|                       Mississippi State University                      |
|_______________________Engineering Research Center_______________________|




More information about the Beowulf mailing list