Agreed Steve and Lonnie. It's all my fault that I am on an ancient
code !
But as they 'once bitten twice shy', I am yet to gather courage to
keep working with XFS.
We have servers working on ext2/ext3 w/o a glitch. That's why I said
unless I see 'silverline on dark clouds' i,e those 'extra' advantages
XFS provide over others, I would like to play safe.
Mahesh
______________________________ Reply Separator _________________________________
Subject: Re: XFs stability
Author: lord (lord@xxxxxxx) at internet
Date: 7/22/2003 6:51 PM
On Tue, 2003-07-22 at 08:04, mahesh.babbar@xxxxxx wrote:
> Thanks for the response Mike & Cliff,
>
> On our server though, which is basically a file server, serving
> clients files over NFS. We have had 4-5 instances of xfs_shutdown
> with message like "in-core memory corruption...." .
>
> The exact messages are:
>
> kerrnel: xfs_force_shutdown(lvm(58,2),0x8) called from line 1035 of
> file xfs_trans.c. Return address = 0xc01cc10a
> kernel: Corruption of in-memory data detected. Shutting down
> filesystem: lvm(58,2)
> kernel: Please umount the filesystem, and rectify the problem(s)
>
> At once, we had lost data even after repairing the filesystem using
> xfs_repair and that too production data. Clearly, we are a worried lot
> !
>
> Would appreciate if I can get some information on concrete advantages
> which XFS has over ext3/2 in terms of stability (specially as a file
> server, performnace (with figures e.g how much % write/read gain) and
> features.
>
> I believe we are using the XFS version which is as old as July 2001.
It is really difficult to have sympathy for you if you are running 2
year old code. There have been a vast number of improvements in the
code since then.
As for advantages/disadvantages, I will leave that to other folks.
Steve
|