File Restore Fails - Legato 5 & 6

[cpeters1]

Please excuse me if I do not provide all info on the first try. I'll try to keep it brief. We have a Legato release 5 server doing backups for a mix of Solaris, Linux and Windows machines. The Clients are running a mix of release 5 & 6 versions. We do nightly incremental backups and roughly monthly full backups. We have 4 drives, three defined as Full and one as NonFull. These are 400GB LTO-2 drives.

I have inherited this system from the previous IT guy who quit, and I personally have little IT experience. I have been trying to restore a file on one of our clients and it always fails. I cleaned the drive, with no luck. However, I noticed I can restore other clients, from the same backup same tape. It seems to be just this one client that has issues. I turned on verify and it does say verified in the completion log. I moved the tape to another drive with the same results (I have been forcing backups to test). I am pretty convinced it is not the tape or the drive, since I can consistently restore other clients from the same tape/drive. I have also been able to restore a file to this client from a full backup with no problem.

The log file show this when I try to restore a file to this client from the incremental tape.

02/25/08 15:09:30 nsrd: pst1:/opt (2/25/08) done reading
02/25/08 15:09:31 nsrd: pst1:/opt (2/25/08) starting read from NonFull.059 of 3 KB
02/25/08 15:09:31 nsrd: media warning: /dev/rmt/Drive4 moving: fsr 2971: I/O error
02/25/08 15:09:32 nsrd: media emergency: could not position NonFull.059 to file 974, record 2974
02/25/08 15:09:32 nsrd: media warning: /dev/rmt/Drive4 reading: I/O error
<snipped repeats>
02/25/08 15:09:32 nsrd: media warning: /dev/rmt/Drive4 reading: I/O error
02/25/08 15:09:32 nsrd: media notice: volume "NonFull.059" on device "/dev/rmt/Drive4": error decoding block, check system
device configuration, disabling forward ...
02/25/08 15:09:32 nsrd: media info: can not read record 2974 of file 974 on vhs tape NonFull.059
02/25/08 15:09:32 nsrd: media warning: /dev/rmt/Drive4 reading: I/O error
<snipped repeats>
02/25/08 15:09:38 nsrd: media warning: /dev/rmt/Drive4 reading: I/O error
02/25/08 15:09:38 nsrd: pst1:/opt (2/25/08) done reading

A restore from the same client, same file, but a full backup worked just fine:

02/26/08 11:00:19 nsrd: pst1:root browsing
02/26/08 11:00:20 nsrd: pst1:<4>/opt (12/07/07) starting read from Full.098 of 3 KB
02/26/08 11:00:56 nsrd: pst1:<4>/opt (12/07/07) done reading 3 KB
02/26/08 11:01:35 nsrd: pst1:root done browsing

I can provide any other info need, please ask.

Thanks for looking.

 

[605]

Actually, if your description of the environment is o.k., then you are working with an unsupported version. You NW server should always have the highest release. Clients could be older - but not vice versa.

NW 6 introduced the new client file index - this is most likely the reason where your problems come from. But there were also important changes to the media index.

To avoid problems let me suggest the following:
- Do not do any file restore (especially not for a NW 6 client).
- Run a save set directory instead and get your files from there.
- Update your NW server. This will cost quite some money as you must buy update enablers. Your versions are not supported any longer.
Destroy your indexes as they are obviously bad.
- Rescan all you media (hope it is a reasonable number).
This takes some time but you will be able to recover in the end.
- Finally update the clients.

 

[cpeters1]

Humm interesting. I found some Legato documentation which seemed to indicate that an older server with a new client was okay, but vice versa was bad. However, I suspected that at least as a start we'd need to get updated on everything, so that is not a surpise. Thanks for the time.

 

[Chapter11]

a particular version of a server can backup a variety of client versions, usually older, but newer clients on a limited basis.

but server 5 and client 6 is not a good idea - too many major architectural changes between those versions.

 

[cpeters1]

I am told this has been working fine for years. I can also recover files from other clients, also version 6, from the same backup session, on the same tape. Wish I knew more about it all. I did find a reference to the error decoding block message and tried to fix the block size at 64 in the st.conf file, but that did not seem to help. And the fact that the full backups for this same server seem to be fine is also weird. I am guessing that the data is on the tape okay, the client just can't seem to read it.

I guess I could prove that by doing what was mentioned above "Run a save set directory instead and get your files from there.", but I have to figure that out first.

 

[a060463xyz]

Try to make a directed recover of this files to a temporary directory on the backupserver. If this works copy the data to the original location.

But NW 5 is no longer supported the sooner you make an update the better for you.