[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: XFS Filesystem problem

On Tue, 29 May 2007, Andreas Grabner wrote:

Am Sonntag, den 27.05.2007, 20:22 +0000 schrieb George N. White III:

I'd want to do some serious testing before using XFS in your environment.
Is there a reason you can't use ext3?
No. I want to test it before using it at customer sites. At one cutomer
the fsck on ext3 takes 2 hours - they are losing money.

  XFS is needed in situations where
you are losing data and or money while systems are down (remote sensing
and other time-critical high-volume data collection, numerical simulation,
video production, etc.).  I'd venture that most heavily used XFS systems
are not using (register-starved) ix86, and are using SCSI, FC, or SAS
storage.  You aren't getting the full benefits of all the testing that
has been done.
George N. White III  <aa056@chebucto.ns.ca>

Does that mean there should not be problems on scsi-storages?

There should not be problems using ide storage, but I don't think xfs has been exercised as completely (e.g., the really exceptional cases) on low-end hardware, so we don't really know if the people who say it works for them were just lucky. I've had to use xfs_repair a few times on SGI machines after a kernel oops or similar rare event, but only once on a PC. In the latter case the system didn't have enough memory for xfs_repair to run, so I really don't know if it would have worked given more than the 256M that was considered generous when the PIII was young.

George N. White III  <aa056@chebucto.ns.ca>

Reply to: