My experience is the following:
Take the set of saveset instances that are on-line (in a tape silo or
on a mounted device).
If that is empty the take the set of off-line instances.
In both cases ignore any instances marked as suspect.
Now within that set choose the saveset instance with the lowest cloneid.
Usually the earlier instances have lower cloneids.
The implications:
For adv_file devices NetWorker is sensible and give the readonly
cloneid a number one less
than the readwrite device, which means that it will choose the
readonly device.
If you make 2 clones (as we do), make sure the one you keep on site
is the first
one made, and the one to go offsite is made second. This is
unfortunate as it
means your offsite clones (which you normally want made as fast as
possible) have
to be created after you make your local ones.
I wish that you could set priorities for locations. I keep two or three
locations and would love to give each a preference for reading.
[Anyone from EMC listening - I think I asked for this some 6 or 7
years ago!]
On 27/04/2007, at 12:20 AM, Brian O'Neill wrote:
Hmm...no responses? Anyone have answers? I though someone would at
least be able to answer about which "clone" is the preferred one on
a restore...
Brian O'Neill wrote:
I just realized a major disadvantage of using our LTO-3 tapes,
which currently fits almost a week's worth of backups on tape -
thats an aweful lot of data to lose if a tape is unreadable...
So, to limit the possibilities of this, I was thinking of simply
making an extra clone of the savesets, and have just a couple
questions.
Right now, the migration from VTL to LTO-3 is handled by a script
that is wrapped around nsrstage, which clones the volumes then
deletes the existing savesets from the media DB. Thus the clones
are the only copies and are used for restores, no issue.
I was thinking of changing this to:
- Split the tapes into two clone pools, "A" and "B"
- Use nsrclone to clone the savesets on the VTL to pool A
- Use nsrstage to clone the savesets on the VTL to pool B, thus
freeing the VTL storage
This would have to be sequential, so it would take twice as long.
Not a big deal for us right now. We could also keep one pool
nearline, and the other offsite.
Does this make sense? Is there a better way?
Also, if a restore is requested, which clone is it likely to
choose to restore from? If I'm doing a non-saveset level restore,
is there a way to specify to use an alternate clone? Or if the
primary clone fails, is a saveset restore my only option?
-Brian
To sign off this list, send email to listserv AT listserv.temple DOT edu
and type "signoff networker" in the body of the email. Please
write to networker-request AT listserv.temple DOT edu if you have any
problems with this list. You can access the archives at http://
listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu
and type "signoff networker" in the body of the email. Please write
to networker-request AT listserv.temple DOT edu if you have any problems
with this list. You can access the archives at http://
listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to networker-request
AT listserv.temple DOT edu if you have any problems with this list. You can access the
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|