Hi,

On 10/9/2006 9:25 PM, Frank wrote:
> Hi,
> 
> I'm probably missing the obvious, but backups are too important not to
> ask the question, even if it is stupid :=)

Quite right. There are no stupid regarding backups...

> Since a few weeks, a try to backup a few clients on the Internet that
> have bad connectivity. That is not really a problem: if the backup
> succeeds then that is fine, and if the backup fails (client not
> connected, network disappears etc.) then it is unfortunate. I did not 
> think anything about this until today, when I noticed that the contents 
> of the tape did not match the contents of the backup database, and hence 
> a restore fails. As I always have been able to restore my files, I 
> assume that this behavior is typical for the new setup with the 
> unreliably connected clients.

Hm... shouldn't.

> Here are the details. I have a volume called DEZE-0015, according to the
> SQL database the following volumes should be on it:
> 
> 
>>+-------+--------------------+---------------------+------+-------+--------+---------------+--------+
>>| JobId | Name               | StartTime           | Type | Level | Files  | 
>>Bytes         | Status |
>>+-------+--------------------+---------------------+------+-------+--------+---------------+--------+
>>| 7,090 | LaptopDocuments    | 2006-10-08 14:01:54 | B    | I     |    428 | 
>>2,529,684,742 | T      |
>>| 7,091 | LaptopOS           | 2006-10-08 16:57:53 | B    | I     |     62 |  
>>  11,696,692 | T      |
>>| 7,092 | LaptopC            | 2006-10-08 16:58:30 | B    | I     |  1,212 |  
>> 101,578,587 | T      |
>>| 7,085 | CorfuOS            | 2006-10-08 17:01:21 | B    | F     | 11,806 | 
>>2,068,391,753 | T      |
>>| 7,086 | TheaOS             | 2006-10-08 17:43:29 | B    | D     |    283 |  
>>  83,681,947 | T      |
>>| 7,093 | Drawbridge         | 2006-10-09 01:05:07 | B    | I     |  1,647 |  
>> 189,754,414 | T      |
>>| 7,094 | Drawbridge-home2   | 2006-10-09 02:05:06 | B    | I     |  6,664 |  
>>  22,163,687 | T      |
>>| 7,095 | Drawbridge-home3   | 2006-10-09 02:08:40 | B    | I     |      2 |  
>>   4,946,568 | T      |
>>| 7,104 | BackupCatalog      | 2006-10-09 07:10:51 | B    | F     |      1 |  
>> 104,853,720 | T      |
>>| 7,105 | SemiDocumentsMieke | 2006-10-09 14:15:30 | B    | I     |      9 |  
>> 313,290,044 | T      |
>>| 7,106 | SemiDocumentsSep   | 2006-10-09 17:11:21 | B    | I     |      5 |  
>>  38,054,192 | T      |
>>+-------+--------------------+---------------------+------+-------+--------+---------------+--------+
> 
> 
> However, if I list the tape using bls, I only find the first 6 backups:

I deleted all the proof that many of the jobs could not be found on this 
tape because...
...
>>09-Oct 20:32 bls: End of Volume at file 7 on device "DDS-3" (/dev/nsa0), 
>>Volume "DEZE-0015"
>>09-Oct 20:32 bls: End of all volumes.
>>#

... after file 7 the tape is marked as EOD.

> 
> 
> This is very strange, because for the remaining jobs I also received a 
> "completion notification" for the other jobs:

For breviity sake I delete this, too. I think we can assume that these 
jobs were saved to the tape in question.
...
> So, according to this log, jobs 7093 and 7094 are safely on volume 
> DEZE-0015, but according to bls they are not.
> 
> Now I'm trying to restore a file from job 7105, once again the log 
> indicates succesfull storage:
> 
> 
... indeed...
> If I try to restore, I get the following and nothing is restored:
> 
> 
>>09-Oct 19:40 drawbridge-dir: Start Restore Job 
>>RestoreFiles.2006-10-09_19.40.21
>>09-Oct 19:40 drawbridge-sd: Ready to read from volume "DEZE-0015" on device 
>>"DDS-3" (/dev/nsa0).
>>09-Oct 19:40 drawbridge-sd: Forward spacing to file:block 13:0.
>>09-Oct 19:42 drawbridge-sd: End of Volume at file 7 on device "DDS-3" 
>>(/dev/nsa0), Volume "DEZE-0015"
>>09-Oct 19:42 drawbridge-sd: End of all volumes.

Yup, EOD after tape file 7.

> Strangely, it tries to find file:block 13:0, which seems incorrect to 
> me, but I do not not know the internals of the backup system, so maybe 
> file numbers can be skipped(?)

No. What happens is that the tape drive is ordered to fast forward to 
tape file mark no. 13, it starts with that task, and ends after file 7 
because there the tape is marked with an EOD mark.

In my opinion, by the way, Bacula should create a big fat warning in thi 
case.

> Anyway, I'm very confused about the state of my backups, and I get the 
> feeling that some things might not be properly backup-ed, but I have no 
> clue what is wrong.

I see two possible reasons:
1. The tape drive is setup incorrectly, especially regarding "Two EOF" 
and related things.
2. Due to some problem, Bacula wrote an "empty" tape file, thus 
accidentially creating an End Of Tape mark on the volume.

> Can someone please enlighten me?

Enlightening is not something you can get quickly ;-)

Anyway, first make sure your tape is setup correctly. Re-run the btape 
tests, and check other volumes similarly to how you worked with 
DEZE-0015 (which was really thorough and detailed).

If this is a tape drive misconfiguration, fix it :-)

If it is something you only see on this one tape, assume it's a unique 
problem. Don't panic, but closely watch your backups until you are sure 
something similar doesn't happen regularly.

If you find this to be a bug in Bacula - i.e., Bacula writes two EOFs 
immediately after each other - see if it's fixed in a more recent 
version, or file a bug report.

Finally, provided there is more data on tape after the (probably) two 
EOF marks making up a false EOT mark there are ways to try to access 
that data, but it's probably not simple.

Arno


> Regards,
> 
> Frank
> 
> 
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Bacula-users mailing list
> Bacula-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/bacula-users

-- 
IT-Service Lehmann                    [EMAIL PROTECTED]
Arno Lehmann                  http://www.its-lehmann.de

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to