Mailinglist Archive: opensuse (2700 mails)

< Previous Next >
Re: [SLE] Disappointed in 9.3
  • From: Darryl Gregorash <raven@xxxxxxxxxxxxx>
  • Date: Sun, 03 Jul 2005 16:48:02 -0600
  • Message-id: <42C86B22.1010203@xxxxxxxxxxxxx>

Carlos E. R. wrote:

The Friday 2005-07-01 at 16:55 -0600, Darryl Gregorash wrote:

I'm about to try another install of 9.3 in the next few days -- maybe I'll
wait until mid-week when I can get some more CDRs, and fetch the 5 install iso
files -- but I doubt I will be using XFS, unless it can be demonstrated
conclusively to me that this won't all just happen again.

Nobody can demonstrate that.

Actually, not as difficult as you think. Just give me a reasonable amount of something more than mere anecdotal "evidence". I'll accept two standard deviations on any meaningful amount of real testing, as should anyone.

But I'm using 9.3 and several partitions types: ext3, reiserfs, xfs... no
problems. I have done forced reboots several times (I'm trying suspend to
disk and it crashes), and I haven't detected corruptions so far.

I certainly didn't say I have a problem in *not* writing to the drive.

Your system might have hardware problems of some kind; for example, bad
memory can corrupt journalled filesystems.

Amazingly, any possible such hardware problem has never affected any Reiser partion I have, consistently destroys one particular XFS partition, and consistenly locks up the system when doing large-volume writes to another particular XFS partition. The partition that is destroyed is subsequently unrecognizable, and the other is so corrupted that, although it can be mounted and read, none of the files actually exist. xfs_repair is of no use with either one. No other XFS partition has ever had more than recoverable errors on it as a result of these crashes.

No hardware error can ever produce such consistent failure to manifest itself with one file system, yet show such consistent presence and behaviour with a different file system. Even if the XFS driver is always loaded into one particular region in memory, and that region happens to have intermittent errors, on balance of probability, such consistency is impossible. As confirmed by significant testing, the drive is OK -- so what else is left?

< Previous Next >