On Mon, 2005-12-05 at 05:24 +1000, Michael Mansour wrote:
..
> Although I have had no issues with running XFS (and you too obviously), I
> presume they have the resources, the clients, the test beds and the know-how
> which far exceeds my own environment and yours. I'd also presume they've seen
> worse disasters in environments than you or I have seen in our own.

Just for the record, we have seen several instances of mysterious FS
corruption under heavy load with XFS on SL(C)3 (e.g. see
http://oss.sgi.com/archives/linux-xfs/2005-06/msg00106.html).

We still use XFS in production.

We believe these bugs to be mostly due to the fact that RHEL3 and hence
SL3 use an old kernel (2.4.21), and the XFS port to that kernel is
pretty much abandoned by the developers. Backporting a newer XFS version
to 2.4.21 turned out to be infeasible.
We hope that things will be better on SL(C)4, at least we will be a lot
closer to where XFS development is happening right now.

Regards
jan