On Sat, 19 Feb 2000, Jonathan Nicholson wrote:
>
> I just had an ADVFS file system go down because it couldn't read the last
> block of a volume in the domain.
>
> The domain is made up of 2 disks rz9c & rz11c. rz9 is a partition of an
> HSZ50 raid 5 set and rz11 is a mirror set off the same hsz50.
>
> Everything is redundant (dual hsz50 dual psu etc) and the disks are all
> mirrored/raid so what could have caused this?
This turned out to be because the disk label on the disk was incorrect.
The disk is presented to 2 nodes in an ASE cluster - the 2 nodes were
seeing different disk labels (presumably a stale entry from a previous
disk on the bus)
Zeroing & Re-labeling the disk (on both nodes!) seems to have fixed the
problem...
Regards,
Jonathan
----------------------------------------------------------------------------
= Jonathan Nicholson - Team Leader : System Support "Special Projects" =
= The Sanger Centre, Wellcome Trust Genome Campus, Hinxton, Cambs, CB10 1SA =
= Email: jjn_at_sanger.ac.uk -=- Tel: 01223 834244 x4987 -=- Fax: 01223 494919 =
----------------------------------------------------------------------------
Received on Mon Feb 21 2000 - 08:54:22 NZDT