Have you verified you have backups from that date?

When you are doing the restore what do you see if you type versions instead of changetime?

Can you explain what the significance of running the mmrecov and scanner commands is?  Have you rebuilt your Networker server?  Did you run nsrck -L7 to recover the indexes after mmrecov?

Is there any kind of break in the savesets or do you have a daily backup from Sept 2 to the present?

There is a much easier way, just restore the indexes for the date you want to recover from:

nsrck -L7 -t "6 October 2010" clientname

Which should make the indexes browsable as per that date (it won't affect any indexes currently in existence, only add to them). If we still appear to have issues after that we could try doing recovery direct from the savesets.  My suspicion is that you are backing up with a one week browse time and a one month retention time though I know that is contrary to what you have initially told us - it would be useful if you could double check how your backups have been set up.

Can you post the mminfo output for the saveset you are recovering,

mminfo -ot -q "clientname=clientname"

It would be useful to see what Networker currently believes is there.

I'm seeing no backups prior to October 6 so the restore is running as expected even if it is not running how we want it to.  If you run the same command on your production server what do you see?

Could you also run:

mminfo -q "savetime < 6 October 2010"

- and see if you get any output (don't specify any client name, I would like to see if there are any backups prior to this time).

I could not see the mminfo output you posted...

If the mminfo command returns the same info on the server returns the same information as on the test recovery server then the issue is not within the rebuild of the server.  However, as the output is showing the same data it suggests that we do not have any backups prior to the 6 October.  You state " Running mminfo -q "savetime < 6 October 2010" shows all of the tapes that have been used in the past" which I take to mean "we can see savesets from other clients prior to 6 October".

The reasons could be down to a few reasons:

* backups failed due to client, network or server issue since resolved

* backups did not run due to configurational issues since corrected

* backups were successful but had a short retention and have since been written over

* backups have been removed by manual activity intentional or accidental

I would suggest the next move would be to grep the Networker logs for that client name and see whether backups were running and successfully completing prior to 7 October or whether or they may not have ran or have had errors.  Once we ascertain this we know a little more about why we are in a situation at present.

Can you just confirm backup-bb is the server we are doing a recovery from?  Is orbapp-1 another client you are doing recoveries from?

We have backups listed below that start 07/02 and go through to 10/04 - do we have a a series of backups in question for the file system we are doing the recovery from?  How about the previous full backup to the 10/04 backup?

Thanks for the info regarding the restore.

I only see that message if I run the 'changetime' command. Otherwise I can restore from the latest backup without any issue. If changetime is within the last week, it's good.

Current working directory is /root/
recover> changetime 10/07/2010
6497:recover: time changed to Thu 07 Oct 2010 11:59:59 PM PDT

Hi,

Let me answer your question first. To recover the index files all you have to do is run "nsrck -L7 -t backup_date client_name".

Secondly, do you see the same message the moment you run the recover command or just when you do a changetime?

HTH,

Rovin D'Souza.