Hi David,
thanks for your answer and also thanks to Matthew and Adam for their hints.To
answer your
questions:
We - better I - use SuSE-Linux 6.0 which contains the (SuSE-)patched kernel
2.0.36 on a SMP-system. The technical data is:
a pair of PII-400, Asus P2B-DS mainboard, 512 MByte of RAM, three Adaptec
SCSI-controllers (yeah, I know, what you are thinking) and
a lot of disks (IBM-type UW).
The disks/filesystems are direct connected to the machine and not exported over
nfs - I think that is what you wanted to know about knfsd....
I had run a fsck, but that did not solve the problem and I don't understand it
until now (we are still searching for the reason). Over all it was
the first time I have seen something like that - normally I am working with
some (really competent) guys on HP-machines and they wondered
about it like I did.
Nethertheless, I will keep on searching.
Thanks again, Dietmar
David Mansfield wrote:
> > At work we use oracle8-database on linux. While benchmarking
> > i/o-performance with stripes (raid 0) and single harddisk, we used to copy
> > the database-files from the stripes to the single harddisks several times.
> > Once I copied the SAME files from the one stripe to a single harddisk and
> > so I would do with the files from another stripe, but it won't work!!!
> > The database was always shut down (really!!!) so that the files were also
> > closed up. All I got was something like "Free blocks count...". I tried it
> > several times using tar and dump but anything works (messages: "I/O
> > error...).
> > Does anybody got similar problems?
>
> Hmmm.. I use Oracle8 also on linux, but NOT the software RAID from Ingo.
> When I got this problem it was an EXT2-fs corruption thing. An fsck of
> the partition was required to repair it.
>
> I wonder what the cause is? What kernel are you using? Is your machine
> SMP (more than 1 CPU)? Are you using knfsd?
>
> David
>
> --
> /==============================\
> | David Mansfield |
> | [EMAIL PROTECTED] |
> \==============================/
