server changed name and problem using older backups

[tekkanet]

I had to change server name for both clients and backup server, not using any more /etc/hosts file but company's internal dns.
After tricking in console and chenge server name to connect to and client name configuration I was able to mantain alla the configuration and backup sessions continued to work.

But now trying to restore older backups for the clients and browsing to a date older than dns change, into the nwrecover session I receive the message

Nothing in index at this time

I can see that inside the media in jukebox indeed there are the relative ssid files.

I also am running the scanner command (not yet finished)

root# scanner -i -S 2280245858 /dev/nst0
scanner: scanning LTO Ultrium-2 tape 000021 on /dev/nst0
scanner: LTO Ultrium-2 tape 000021 already exists in the media index

and I'm waitying it to finish, but I presume it will change nothing...

Any hints to have the nwrecover session take benefit of what contained inside the tape...?

server and client are Linux with 7.3.2

Thanks in advance,
gianluca

 

[605]

You should always use scanner along with "-v" to receive a kind of "heartbeat" as well.

The problem you might have is that you have renamed/deleted the CFI directory for the old client (server) so that NW can not find it any more. You should be able to solve the problem as follows:
- Add an entry for the old machine in the hosts file so that Networker can resolve the hostname
- Recreate the old client
- Recover its old CFI
- Use the directed recovery for restores

 

[tekkanet]

probably I didn't explain correctly.
Both the client and the server didn't change at all.
They only was renamed from a dns domani point of view.
Suppose that before they both had in /etc/hosts the entries:

192.168.0.18 bcksrv.company.com bcksrv
192.168.0.2 bckclnt.company.com bckclnt

and the server knew itself as bcksrv, while the client as bckclnt.

When named changed I was unable to connect to server from java console so that I hado "add" a new server in it (bcksrv.dept1.company.com : I added a subdomain part), but actually it was the same.
Infact after connecting, I saw all the previous configuration.
I also needed to change client parameters as I was not able to connect to them.
And I began taking backups as usual.
Now I need to recover by previous sessions and I get that error about media index....

Also I didn't understan what you meant with CFI and with the terms "directed recovery".

I tried also from another client to use the -c and/or -s switches with many mixed values for both client and server names, but without luck.
When able to start X gui I get the same error, in other cases RPC error without being able to open the gui....

 

[tekkanet]

to be more precise about name change, after setup of internal dns server I commented out the lines in /etc/hosts and now both of them are able to see each other with the full name and with the short name.
Before they were a sort of
host.company.com through /etc/hosts
now they are
host.dpt1.company.com throuch dns
thanks in advance

 

[605]

CFI = Client File Index

Directed Recovery = Well known term to recover from one client to another. (sorry, for Linux only available from the command line, but you may start this from the Windows client if you have such a client)

The problem you have is obviously due to the fact that you created the clients with the short names. If you would have done it using the FQDN, NW would have added the short name to the client's Alias list as well. I assume, the FQDN is now missing and you should ensure that the CFI can be used by the long and the short name.

 

[tekkanet]

It seems that after the scanner command (I interrupted it for evening backups and I reran it with the -v option then) I solved my problem.
I mounted the tape for a session I needed (and that was before change of name event) and ran

scanner -i -v -S 2280245858 /dev/nst0

...
scanner: scanning file 278, record 31000
scanner: scanning file 278, record 31100
scanner: scanning file 278, record 31200
scanner: scanning file 278, record 31300
scanner: scanning file 278, record 31400
scanner: ssid 2280245858: scan complete
scanner: ssid 2280245858: 28 GB, 51375 file(s)
scanner: (ssid 2280245858) added 16407 new file index entries (34968 already existed)

At this point, connecting on a client enabled to do recover for the original one I could run

nwrecover -c bckclnt

and I was able to "Change browse time" at 4th of March without receiving anymore the error message about "nothing in index at this time" and I could successfully complete my recovery.

Thanks for hints and pointers.
Gianluca