> # Your choice of low-level controller driver for the attached hard
> # drives can greatly
> # impact this problem. I had a machine with an IDE drive and a Triton II
> # chipset. When I enabled the Triton DMA support, I would get corruption,
> # reliably, and under light load even. When I disabled the special drivers,
> # the corruption went away.
> Granted, ext2 could probably use better error handling, but there's
> bound to be more to it than a resounding "ext2 sucks".
This is certainly true. There are some very large sites running Linux
who don't seem to be having any problems. It would be good to know what
type of controllers are being used by the people having problems. It
would also be extreamly helpful to have a reproducable way of creating
this problem that was simpler than running news for a few days.
My $.02
Jim