On 01/07/2013, at 17:47 , bingo <networker-forum AT BACKUPCENTRAL DOT COM>
wrote:
> @tony
>
> Out of my experience your statement:
> "And as a side note, remember that a clone is exactly that, a clone. This
> means that NW doesn't care the least whether you point it to a clone to do
> the recover from or the original saveset."
> is not really correct. At least not when NW selects the recover media.
>
> NW will in fact always go for the original backup instance - precisely the
> one with the lowest clone ID timestamp.
Correct AND incorrect.
NetWorker will *always* go for the backup instance with the lowest clone ID.
While this is most likely the original backup, it isn't guaranteed.
E.g.,
Backup to Tape
Clone to Tape
Stage to Tape
In this scenario, the "Clone" copy will get requested ahead of the Stage (after
the original backup is removed), even though technically the staged version is
likely on a backup rather than backup clone tape.
> This makes sense as otherwise he might potentially grab the slower tape. It
> is a pain if you set the location for a backup tape media to another value
> than your jukebox' name as NW will not even choose another media - even if it
> is inside the autochanger - unless you will force him to do so. And you can
> achieve that in several ways:
> - mark the original backup save set 'suspect'
> - for save set restores: specifiy the save set instance (-S ssid/cloneid)
Neither of these options are particularly desirable.
Instead, make use of the offsite facility, introduced in NetWorker 7.3:
nsrmm -o offsite volumeName
NetWorker will not request a volume that has been flagged "offsite" even if the
SSID on it has the lowest clone ID if there is another copy on any volume that
_isn't_ marked "offsite".
Cheers,
Preston.
|