On Mon, Nov 08, 2004 at 10:50:29PM +0100, Michael Tautschnig wrote:On Mon, Nov 08, 2004 at 10:38:10PM +0100, Michael Tautschnig wrote:Ok, I will try - but what is the volumeid (i.e., where can I find it?)If 'volutil getvolumelist' lists a volume, then the volume exists and should be purgeable with 'volutil purge <volumeid> <replicaname>'.
Both the volumeid and the replicaname do have to match. And the replica name is not the same as the replicated volume name, but has a '.0', or '.1' appended.
# volutil getvolumelist W<volumename> I<volumeid> H1 ...
Thanks - and that does the trick. Some debugging shows:
volumes=`echo "$volumelist" | awk '$8 ~/^W'$REP'$/ { print $2 }'`
in purgevol_rep returns an emptyp "$volumes" - although that command is called. Could you check the commands before that line? Actually $REP contains the volume-id (not the name!)- is that correct???
That should be correct, the 8th field in the volumelist has the id of the read-write copy the volume was derived from, it is identical to the volume id on the r/w replica itself. This way we can find any backups or clones that belong to the original r/w replica.
Wvmm:u.jaharkes.1 Ic900007a Hc9 P/vicepa m0 M0 U3c01 Wc900007a C41386f7a D41386f7a B4190596b Aca6 ^^^^^^^^^ Bvmm:u.jaharkes.1.backup Ic90001c5 Hc9 P/vicepa m0 M0 U3bc0 Wc900007a C413a597c D413a597c B0 A0
Of course, you are right - but the problem here is: my volinfo reads as follows
...
Replica0 id 0100004a, Server0 192.168.248.3
...
i.e, with a leading "0" , whereas getvolumelist prints:
Wdeguilla.0 I100004a H1 P/codaserver m0 M0 U2 W100004a C417386af D417386af B0 A32
!
would it be okay, if we simply removed any leading zeros?
Thanks, Michael
