[Dirvish] Bypassing --init to re-use an existing copy

peter.billam at dpiwe.tas.gov.au peter.billam at dpiwe.tas.gov.au
Thu Mar 2 19:59:04 EST 2006

Jason Boxman wrote:

> Or you could hack the vault itself:
> sarah:/snapshot/edseek/edseek-web/dirvish# cat default.hist
> 20060209        2006-02-09 03:52:42     default +30 days == 2006-03-11 
> 03:38:36
> echo "Status: success" > /snapshot/edseek/edseek-web/20060209/summary
> Then move the files under your /bank/vault/YYMMDD/tree.
> I believe the summary file is the authoritative file when it comes
> to whether the backup completed or not.

Hmm, as far as I can see, that's exactly what I tried yesterday,
and the disk started filling up...  So I'll try the cp -al option.

> One hack might be to cp -al the current link or
> rsync --link-dest=/foo /foo /bar
> and then do an initial Dirvish backup of that, then change your
> default.conf to point to the true target machine for future backups

I'm unsure why the initial cp -al is necessary... can I be more specific ?
The backup machine is terry2 and the primary machine is terry.
On terry2, the data currently resides in /backup/terry/
It got there by  rsync -aq -e /usr/bin/ssh terry:/data/terry/ /backup/terry
The vault is /backup/terry_data/

So do you mean:
# cp -al /backup/terry /backup/terry_cp
# hostname
# egrep 'client|tree' /backup/terry_data/dirvish/default.conf
client:         terry2
tree:           /backup/terry_cp
# dirvish --init --vault terry_data
# vi /backup/terry_data/dirvish/default.conf
# egrep 'client|tree' /backup/terry_data/dirvish/default.conf
client:         dirvish at terry
tree:           /data/terry

Is that right ?  Why the cp -al stage ?

Thanks for your help,  Regards,  Peter

AUS/TAS/DPIWE/CIT/Servers hbt/lnd/l8   6233 3061   http://www.pjb.com.au
  I met a man whose name was Time; he said "I must be going..."
  But just how long ago that was I have no way of knowing.
  - October Song,  1st LP,  Incredible String Band

More information about the Dirvish mailing list