Breaking up RAIDSETs

From: <peter.mcdonald_at_afp.gov.au>
Date: Fri, 06 Mar 1998 11:32:39 +1000

People:

During our scheduled downtime last night we planned to break up a 20 Gb RAID 3/5
raidset, to steal two of the spindles to increase the available space for our
Oracle 7.3.3 archive logs.

We run a two node 4100 cluster, with 1 Gb RAM in each, DUNIX 4.0B/wPatch 6, TCR
1.4a, AdvFs, and NSR on one node.

The AdvFs volumes on the target node were unmounted, and the data copied to
another volume with lots of space. We then used HSZterm (Virtual Terminal mode)
to access the HSZ50 dual-redundant RAID controllers.

We issued the command:

delete d204

D204 being the container name of the RAIDSET to be broken up. The process then
hung, and went uninterupptible. We rebooted the machine, and after some initial
fright, recovered the AdvFs volumes from the other node (they had failed over).

I have the uncomfortable feeling we missed something out, but my reading of the
documentation doesn't suggest that.


Can anyone provide any light on the subject?


Peter McDonald
Australian Federal Police

peter.mcdonald_at_afp.gov.au
Received on Fri Mar 06 1998 - 07:15:04 NZDT

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:37 NZDT