Sorry for the goofy format; my new Lotus client is being obstinate about
using plain text only ...
>If the old server's client id does not match the new server's client id,
then step 3 will not work. Reason is simple. NetWorker will expect the
index you want is owned by the client id of the current >server, and
savesets are owned by the current server's client id.
MJL:
Yes, I know. But the clientID of the current client "Old-ADMNCRY001" does
match the clientID of the old client "ADMNCRY001". I created the current
client "Old-ADMNCRY001" with the clientID of "ADMNCRY001" from 2 years
ago. So the clientIDs do match.
>My advise is for you to leave your current production server alone. Using
the same process for a typical disaster recovery of the NetWorker server,
create a new NetWorker server using the same name as the old >one. To
avoid a network conflict, either use the local host file to assign it an
unused i.p. address, or disconnect it off the network.
>Make sure that name resolution works the same way. If the old server was
known by it's short hostname, then in the local host file put the hostname
right after the i.p.address. If the old server was known by >it's fully
qualified domain name, then put that right after the i.p. address. This
is critical if you want NetWorker to use the same client id.
>Once this is done, then connect a compatible tape drive, and then recover
the bootstrap, then recover the index. Voila, you now have a NetWorker
server that has all the information prior to the server >crashing. You
can now access the media database, and find out where all the backup
information is stored.
After 2 years, I don't remember if it was known as short name or FQDN. :-)
I certainly remember the name of the server, and can set up a machine with
that same name. It's been my experience (with 1 NW Network Edition server,
and 4 Business Editions) that servers show up as FQDN, in the client list.
So I'm guessing that the installation will create it as FQDN.
>>> Scanner -im: since these are foreign Tape to NW server. Does the order
matter? I am not Sure with the New versions I have not done this with
random Tapes. But in older version of NW the order of the tapes really
matters.
>
>> My reply: Yes, the order of scanning is (usually) important - if the
savesets span multiple tapes. But I am *not* running scanner here, am I? I
want to run nsrck -L7 to restore the bootstrap (i.e., index values) for a
specific client. Isn't that different from "scanner -i"? I don't know
which of 20 tapes has the info I want, and I'm not about to run "scanner"
on all of them, to find which volume I need, if I have the index available
as a bootstrap. Isn't that what the bootstrap is for - avoiding using
"scanner -i"?
>Absolutely TRUE. Especially because savesets can span multiple volumes.
Without the media database, there is no way to know what order the volumes
were used unless you first scan in all the volumes, then >query the media
database to determine the order used, then scan the tapes AGAIN in the
correct order.
MJL:
Yes, I know. I've had to do that several times - do "scanner -m" to
populate the media db, then a mminfo showing the "fragflags" attribute, to
get the order of the volumes, then a "scanner -i" to rebuild the indexes
in that order.
>> Actually, the client is already in the mm database. It got in there
when I did "scanner -i" on some older tapes, when recovering for a
different older client( I got entries for a lot of old clients that way,
all named
>> "~client-name-1", since there already was a "client-name"). So I
created a client with the same clientID as the client had on the old
server, and now it shows up as the "Old-client" name, instead of
"~client-name-1".
>"~client-name-1" shows up because the client id does not match the client
id that is currently defined in NetWorker. Similarly you can see
~volume_name after a scan if the volume id of the tape does not match the
one in the current media database.
My reply:
Yes, I know. I have 2 other client entries this way, that I have used for
recovery via "scanner -i" on old tapes. The difference between those times
and this .. is that I want to do "nsrck -L7 -c client" to avoid running
"scanner -i" on a bunch of tapes.
>> And on the bootstrap tape I will get tomorrow, are entries for a client
with the same clientID as my current client "Old-ADMNCRY001" has. So all I
am really doing is restoring the bootstrap for what NW thinks is a valid
client. Even if there were no entries in the mm database for
"Old-ADMNCRY001", it should just create them, since it will find entries
on the tape for a clientID that matches that client definition. That's
been my experience, anyway, as I have successfully done this same
procedure of creating clients with the old clientID previously, and then
used "scanner -i" to populate the index. It's just that this time, I won't
be using "scanner -i", but "nsrck L7 -c client".
>Ohhhh Boy... I think you are under the impression that recovering the
bootstrap will merge the media database from backups with the one that is
online?
MJL:
Yah, that's what I'm hoping ...
>Well IT DOESN'T. nsrck -L7 merges index databases ok, but recovering the
bootstrap replaces the current media database with what you just
recovered. And I'm sure you don't want to lose the information in >your
current backup server. This is the reason what I really think you should
be doing this on test machine.
My reply:
Ratz. :-( Well, that stinks. You'd think that by recovering the bootstrap
for that one client would populate the index db only for that one client.
So you're saying that even doing "nsrck -L7 -c client" for *one* client,
overwrites my *entire* current index db?? I'm not recovering the whole mm
db with "mmrecov", nor importing the entire bootstrap with "nsrck -L7". So
why would it replace the whole media and index db??
I have no old hardware to use for a temporary server. What I have, is a
storage node with that old style tape drive (SDLT) that I use for
recovers. I'd have to delete it; rename it as the old server name; install
a trial copy of NW; do a full D/R recovery; recover my 1 or 2 files ...
then delete the whole thing; rename it back as it's current storage node
name; and re-install it as a storage node in my current server ...
If I do all that, can I use NW 7.5.2 as the (temporary) server? I should
get a 30 day grace period for activation ...
Only other way I can think of .. is to do "scanner -m" on all the tapes
that client might be on (there are about 20, from the EOM run I need),
then do a mminfo showing which volumes and which order I need for my one
client, then do a "scanner -i" on those specific volumes, *then* do the
recovery.
Correct?
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
|