Topic: Restore of a Backup does not work as expected

[apaeffgen]

I use the standard backup funktion of SMEServer 9.2 (Latest updates)
1. Backup to Workstation
2. CIFS Protokoll
3. Multiple Servers in one directory
4. Keep up to 7 sets of the backups

The directory structure begins, like in the documentation stated with the hostname
After that, there is destinctive name, that is not used in SMEServer. I do not know where this comes from.
And in this directory all backupsets are located with the names like 2019.09.13.8.dar. The first is the date stamp. yyyy.mm.dd + the number of the set. In this case it is the last file of the set 2019.09.13.x.dar

When i want to restore the last set the server-manager gives me only one choice of a backup, called 2019(Domain),  mounts the correct cifs share but does state there is a problem restoring the files.
My question now is:

How do i have to rename / restructure the backup-directory, so the server-manager of SMEServer can restore the last / latest backup?
Or is there another solution to the problem.

[janet]

apaeffgen

When i want to restore the last set the server-manager gives me only one choice of a backup, called 2019(Domain),  mounts the correct cifs share but does state there is a problem restoring the files.

Please provide the error message

Note that the restore should occur from the same location as stated in the backup configuration, so there should be nothing for you to have to change.

[apaeffgen]

Unfortunately the error message in the gui is in german:
Es ist nicht möglich, mit der Wiederherstellung der Servereinstellungen fortzufahren.
Wiederherstellung wird abgebrochen, da die benötigte Gesamtsicherung fehlt oder nicht lesbar ist.
Translates into:
It is not possible to recover the server settings.
Recovery canceled, because the complete backup is missing or not readable.
If i check the last file of the actual set with dar, everything is there and readable.
Maybe i missed something in the message logs?

[mmccarn]

I added a test server to my existing server backup share. My backup share name is "smeserver".  If I open \\ip-of-backup-server\smeserver in 'Finder' on my mac I see this folder structure:

smeserver
  - office.realdomain.tld
    dar-catalog
    set4
    set5
    set6
    set1
    set2
    set3
  - sme92.testing.local
    dar-catalog
    set1


Each "set#" is a folder containing one or more ".dar" files.

'office.realdomain.tld/set3', for example, contains 64GB of data in 94 "full-..." files:
  full-20190908164226.1.dar
  full-20190908164226.2.dar
  full-20190908164226.3.dar
  ...
  full-20190908164226.93.dar
  full-20190908164226.94.dar

Plus 6 "inc-" files:
  inc-001-20190909164226.1.dar
  inc-002-20190910164226.1.dar
  inc-003-20190911164225.1.dar
  inc-004-20190912164225.1.dar
  inc-005-20190913164228.1.dar
  inc-006-20190914164226.1.dar

The new test backup set folder looks like this:
  full-20190915093603.1.dar
  inc-001-20190915095802.1.dar

To create inc-001 for the test server, I reconfigured the backup time and some other settings.

I have confirmed that if I execute the workstation restore and select the "inc-001..." file, I end up with the final backup time (9:58am).  If I go back and run the restore using the "full-..." file, I get the original backup time (9:36am).  The restore summary shows the same number of files after each restore (but I didn't add any files, so that's to be expected).

If your folder structure does not look like what I've shown there may be a problem outside of the workstation backup system.

Notes:
* It appears to me that for 'restore' to work on a new or re-installed server you would have to give the new server the exact same name as the old/failed server, then configure (possibly without enabling?) the same settings in 'Configure workstation backup'.

* During the post-restore reboot, the system checks for IP address conflicts -- I don't know what happens if the source server is still online.

* I was unable to find any useful logging information related to the backup on either my live server or on the test server. 

* I started a "Verify workstation backup" job on my live server.  That job has now been running for 65 minutes -- and I don't think I'm going to get to see the results since I accidentally navigated away from the server-manager page and I can't find any dar-related log files anywhere.  I know it's still running because:

Code: [Select]

ps auxwww |grep dar
root     23384 96.5  7.0 318156 285856 ?       DL   09:39  57:14 /usr/bin/dar -Q --test /mnt/smb/office.realdomain.tld/set3/full-20190908164226 --noconf
root     28908  0.0  0.0 103324   860 pts/0    S+   10:38   0:00 grep dar


[janet]

apaeffgen

Error message:
"It is not possible to recover the server settings.
Recovery canceled, because the complete backup is missing or not readable."

It says "missing or not readable".

Note that a full restore should only be done to a new cleanly installed version of SME server.

You should NOT do a restore to an existing configured server that has existing data, as serious inconsistencies can occur.
There is an option to do file only type restores if you are trying to recover a file or a few files.

In the case of a full restore:
On the new "blank data" server you should configure the backup job identically to the old server so that the correct backup location is accessed to perform the full restore.
Basic configuration eg Server names & domain name should be identical (ie between the old & new servers), & the old server should be taken off the network.

Can you confirm you are doing the above steps correctly ?