Dear admins,
I wrote:
last night one member of a ASE-cluster crashed. The remaining node tried to
start the service (Oracle database) and failed (reasons still not clear).
Hosts: 2x Alpha 2100, DU 4.0B with patch 7, ASE 1.41
Right now we have both members up and the diskgroup is deported, all disks
offline.
I can set the disks online and can read the configuration with voldisk list
<disk>. Everythinks seems fine, disklabels are ok, too.
However, voldg import dgname AND voldg -f import dgname failed with "Disk group
has no valid configuration copies".
Digital CSG has no further ideas and we're running out of time. Do you have any
ideas how to import that diskgroup?
We'll start to restore the backup soon, but that is going to take some time...
I recived on reply (thanks to Felix Maurer).
The solution came from Digital CSC:
LSM Metadata of that diskgroup were corrupt.
Fortunately, we run volsave each night and the metadata is stored by that tool
(.../dgname.d/dgname.DF)
We created a new diskgroup named dg2 with an empty disk, added all disks from
the original diskgroup and restored the metadata.
This procedure is described in the LSM Manual, Section C.13.
In short:
# voldg init dg2 rze84=rze84
# voldisk online rza81
# voldisk online rz...
...
# voldg -g dg2 adddisk rza81=rza81
# voldg -g dg2 adddisk rz...
....
# volmake -g dg2 -d p11dbdg.DF
After that, all volumes must be startet and the Advfs-Domains have to be
created by hand
(Modify existing links or create dir in /etc/fdmns and create links to volumes
in /dev/vol )
Oracle is up and running again!
Peter
Peter.Braack_at_degussa.de
Received on Wed Sep 09 1998 - 09:54:46 NZST