Re: [Bacula-users] Listing jobs and volumes...

2006-11-23 Thread Georg Altmann


--On Donnerstag, 23. November 2006 17:36 + Jaime Ventura 
<[EMAIL PROTECTED]> wrote:

> Hello,
> Is it possible to get a "list jobs" having also listed the volume(s)
> related to them?

This is in the manual:


Use
list jobmedia jobid=

Regards,
Georg


-
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


Re: [Bacula-users] advance required volumes

2006-11-24 Thread Georg Altmann
--On Freitag, 24. November 2006 12:13 + Bill Hill <[EMAIL PROTECTED]> 
wrote:

> Sorry if this is obvious, but I can't see it anywhere.
>
> I'm using tape drives that are not in an autochanger. How
> can I find out (preferably be email) what volumes are
> required for the coming nights backup?

My suggestion would be to make a cron job. Something like

echo list volumes | bconsole | mail -s "Volume Listing" 
[EMAIL PROTECTED]

Then you can choose which volumes to use. Any writable (i.e. not {full, 
readonly, etc.}) volumes from the correct pool will do fine. Depending on 
your setup, you could query only a specific pool: list volumes pool=Full

Regards,
Georg

-
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


Re: [Bacula-users] Backing up an entire disk partition?

2006-11-25 Thread Georg Altmann


--On Samstag, 25. November 2006 04:41 -1000 Hydro Meteor 
<[EMAIL PROTECTED]> wrote:

> Oops, looks like I was able to answer my own question for the most part
> as I had missed parts in the current manual describing configuring to
> back up a raw partition (sparse = yes). Even so, it would be interesting
> to hear from anyone who has some experiences and scar tissue from doing
> so in a production environment.
>
> Even more specifically, besides flavors of Linux, it would be interesting
> to hear of anyone who has any direct experiences with HFSX (the version
> of the HFS+ Apple file system that is case sensitive). I can't imagine
> case sensitivity being a problem since Apple's HFS+ deviates from most of
> the world's modern day file systems which are typically case sensitive.

AFAIK the fs-type is not of much interest when backing up a raw partition. 
The partition/fs should not be mounted though for obvious reasons.




> On 11/25/06, Hydro Meteor <[EMAIL PROTECTED]> wrote:
>
> Hopefully this doesn't come across as a question with an obvious answer,
> but I'm wondering if there is any reason why one can not or should not
> use Bacula to back up (and restore later if necessary such as if there is
> a physical disk failure) an entire disk partition of a platform (at the
> moment specifically I'm thinking of a machine or machines with Ubuntu
> Linux installed -- Dapper Drake or Edgy -- but later would want to extend
> to Mac OS X and FreeBSD if feasible)?
>
> The reason I ask is that I've noticed that Bacula hasn't gotten too much
> attention it seems on the Ubuntu wiki with regard to backup and recovery
> solutions and I have seen several on various mailing lists suggest using
> a lower level disk imaging technology for backing up entire partitions
> such as "parted" and "partedimage" of which there are also graphical
> front ends (such as QParted).
>
> Has anyone had experiences backing up entire partitions of any platform
> (be it Linux (Ubuntu or not), Mac OS X, FreeBSD, et al)?

I have never done raw backups with bacula, so I can't tell you much about 
using this for production. IMHO backing up at device level doesn't seem 
very logical for backup and recovery though, if you have the option of a 
direct fs-backup, which is the case for Linux, FreeBSD and OSX.

Regards,
Georg

-
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


Re: [Bacula-users] Error writing the first tape block after changing tape

2006-11-27 Thread Georg Altmann


--On Montag, 27. November 2006 12:30 +0100 Ferdinando Pasqualetti 
<[EMAIL PROTECTED]> wrote:

> Hello everybody,
>
> I am experiencing a frequent error (almost stable, but sometimes it
> worked). Bacula is MySQL version 1.38.11-3 on a RedHat ES 4.04 running on
> an HP Proliant Server. The tape is an HP MSL6000 LTO-3 device with
> autochanger and 2 drives.
>
> When an EOT is reached the tape is unloaded and a new tape loaded, the
> label is recognized correctly, but the first write receive the same error
> response of the exhausted tape, and also the number of blocks is the one
> of the old tape. So the job is aborted. The next job uses the newly
> mounted tape correctly, up to the new tape change and the problem
> presents  again.
> Here are the two relevant portion of logs:

Did you run the btape tests successfully?
http://www.bacula.org/rel-manual/Testing_Your_Tape_Drive.html#_ChapterStart27

Regards,
Georg



-
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


Re: [Bacula-users] bacula restore and cd command

2006-11-29 Thread Georg Altmann


--On Mittwoch, 29. November 2006 09:41 +0100 Pavel Stránský 
<[EMAIL PROTECTED]> wrote:

> Hallo,
>
> when I want to restore some files and trough the tree I come to the
> directory "---=== POSTBOX ===---" I cannot cd to this directory.
> Bacula always says "Invalid path given".
> Of course I am using comma in directory name.

Have you tried
cd "---=== POSTBOX ===---"
or
cd ---\ POSTBOX\ ===---
?


Regards,
Georg

-
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


Re: [Bacula-users] checksums

2006-11-30 Thread Georg Altmann

--On Mittwoch, 29. November 2006 09:07 -0800 Robert Nelson 
<[EMAIL PROTECTED]> wrote:

> You could always do the checksum and copy before the backup.
>
> Just dump the files to a different directory.  Then copy the files that
> have changed to the directory that gets backed up.

Good point. My suggestion would be to use md5 (or sha1 if you're paranoid) 
and store the result along with the dumps. Compare the hashsums before 
backup, and if they are different mv the new dump over the old one.

If your dumps are are textfiles - like mysql for example - you could even 
experiment with diff and patch. So you would have some sort of differential 
backup. Real DBMS support differential dumps anyway... ;-)

Regards,
Georg

-
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


Re: [Bacula-users] Acronis True Image server?

2006-12-06 Thread Georg Altmann



--On Mittwoch, 6. Dezember 2006 09:59 -0600 "Jeremy C. Reed" 
<[EMAIL PROTECTED]> wrote:

> I have a customer interested in using Acronis True Image Server instead
> of  Bacula. Some reason were the Acronis bare metal restore and that the
> Acronis Linux boot CD supports several raid drivers.

I guess there ist no problem creating a linux boot cd-rom having the raid 
drivers you want, is there?

> The website is at
> http://www.acronis.com/enterprise/products/ATISLin/
>
> I don't know the pricing nor number of licenses needed.
>
> The website above says "The only Linux disk imaging and bare-metal
> restore  solution on the market!"

yes and they offer "Peace of mind" by "ensuring that images can be used for 
restoration"... SCNR


I don't know it and website only has the usual marketing blabla. Why would 
you want to use a disk imaging software for backups?
I've been working with 3 commercial backup-"solutions" before bacula and I 
wouldn't want to go back using any of them.

Regards,
Georg


-
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


Re: [Bacula-users] Autochanger Multitape Drives - Unknown Pool

2006-12-06 Thread Georg Altmann

First, I think your mail-client is quoting incorrectly which is very 
confusing. Please consider fxing this.

--On Mittwoch, 6. Dezember 2006 13:59 -0300 Heitor Medrado de Faria 
<[EMAIL PROTECTED]> wrote:

>
> Well, I am sorry but I don't see any error.  It isn't an error if the
> word unkinown is printed.  For a restore, it doesn't care what pool is
> involved.
>
>  > But how can I do my backups, if bacula dont recognize the volume pool?
>
>
>  >> Sorry, Kern. The problem is not in fact the unknown pool. What is
> happening is the Bacula doesnt start writing the second drive volume,
> when the first gets full.
> The first drive stays BLOCKED - Waiting for Media. Any ideas?

Second, I think you got the wrong idea of the autochanger resource. You 
want bacula to continue the job on another drive.
This is not the way autochangers work (not in bacula at least). Bacula 
expects the next volume in the _same_ drive. AFAIK there is currently no 
way to tell bacula to continue the job on another device. Correct me if I'm 
wrong..

Regards,
Georg

-
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


Re: [Bacula-users] Restorejob bigger than space on originalserver (no delete record) - possible workaround but...

2006-12-06 Thread Georg Altmann


--On Mittwoch, 6. Dezember 2006 09:42 +0100 User100 <[EMAIL PROTECTED]> 
wrote:

>> -Ursprüngliche Nachricht-
>> Von: Alan Brown [mailto:[EMAIL PROTECTED]
>> Gesendet: Dienstag, 05. Dezember 2006 13:42
>> An: User100
>> Cc: bacula-users@lists.sourceforge.net
>> Betreff: Re: [Bacula-users] Restorejob bigger than space on
>> originalserver (no delete record) - possible workaround but...
>>
>>
>>
>> > So in theory when creating a filelist we have just to restore the
>> > filelist.txt from the last incremental set and make the
>> full restore based
>> > on this filelist with option  "7: Enter a list of files to
>> restore" and
>> > enter "> "long" with
>> > entries >1 and "useless long" with many entries >10
>> here. Does
>> > anybody have an idea how to workaround another way or to
>> speed this up
>> > great?
>>
>> This is exactly what's been discussed here over the last
>> couple of weeks
>> in terms of point-in-time restores of large backups (which is
>> functionally
>> similar to a migration job and almost identical to a verify job)
>>
>> AB
>>
>
> Thanks for your information. I made a look into:
> http://www.bacula.org/dev-manual/Migration.html but I´m not sure how it
> could help to move data from one volume to another and it seems to me a
> little bit like another "workaround"? I´m afraid this took more
> additional backupspace than a simple filelist which was included on the
> incremental backups and isn´t there the same problem that no
> delete-record is done? If I made on fulljob in the past and have some
> incremental sets from one client how should the migration job know that
> he should just include files that current exist on the client so I don´t
> risk to fill up my server on a disaster? I think if I need to restore
> just one folder which was deleted by a busy user than double-names should
> not be so critical but if I need to restore a crashed server than normaly
> I don´t care about "lost files" but need to restore the last known
> working state (and no duplicates which could avoid this step by filling
> up my harddrives).
>

Um, I think Alan meant the thread "Incremental backup, not accurate?"
You can find it here:
http://gmane.org/

also currently project no. 3:


regards,
Georg


-
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


Re: [Bacula-users] Retension of diskbased volumes

2006-12-06 Thread Georg Altmann


--On Montag, 4. Dezember 2006 16:17 +0100 Marcus Hallberg 
<[EMAIL PROTECTED]> wrote:

> Hi!
>
> I just wants to see if I have understood this correctly... I am backing
> up to disk and I want to configure my retention times correctly.
>
> What I do is to take one fullbackup every 3 months and daily
> incrementals in between.
> What I want is to be able to restore data that is three months old and
> have bacula to recycle my volumes and prune my database. In the example
> I have added one extra month to be on the safe side.
>
consider using a shorter full interval. 3 months full backups will give you 
roughly 90 incremental backups. If you loose one of these backups you might 
end up loosing data 3 month old (in the worst case).
You might also want to read the thread "Incremental backup, not accurate?" 
for other reasons why a long period of only incremental backups is possibly 
not a good idea. You can find the posts here:
http://gmane.org/


Regards,
Georg



-
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


Re: [Bacula-users] Acronis True Image server?

2006-12-06 Thread Georg Altmann


--On Mittwoch, 6. Dezember 2006 19:13 +0100 Kern Sibbald <[EMAIL PROTECTED]> 
wrote:

> In case anyone on this list doesn't understand the difference between
> backup  programs and disk imaging programs (such as Acronis), backup
> programs backup  file by file.  They can restore a full system to the bit
> level, but the bits  may be in slightly different places on the disk.  A
> disk image system makes a  bit for bit copy of your disk.
>
> On Wednesday 06 December 2006 17:54, Georg Altmann wrote:
>>
>> --On Mittwoch, 6. Dezember 2006 09:59 -0600 "Jeremy C. Reed"
>> <[EMAIL PROTECTED]> wrote:
>>
>> > I have a customer interested in using Acronis True Image Server instead
>> > of  Bacula. Some reason were the Acronis bare metal restore and that
>> > the Acronis Linux boot CD supports several raid drivers.
>>
>> I guess there ist no problem creating a linux boot cd-rom having the
>> raid  drivers you want, is there?
>
> I have a hard time remembering when support for raid went in, if I am not
> mistaken the old 1.38.11 Bacula rescue supported raid.

I wasn't refering to the bacula rescue cd-rom. I just wanted to say, that 
it is not a big deal to put the correct drivers for your raid system on a 
linux cd-rom. Or you could just use a well maintained OS like FreeBSD and 
avoid the linux driver/distribution mess at all... ;-)

> Also, what if you want to restore a file to a different system or OS, or
> you  want to restore to a different disk with a different size (perhaps
> smaller  than the original???).  I would have a lot of concerns about a
> disk imaging  system other than taking a disaster recovery snapshot.

You missed
"Acronis Universal Restore — Restore to different hardware and recover 
from any unforeseen event (note: requires separate add-on license)".

I especially like what comes after "note:". :-)

>> I've been working with 3 commercial backup-"solutions" before bacula and
>> I  wouldn't want to go back using any of them.
>
> Thanks.
>
> If we can just get a good complete graphical interface written, we'll
> really  be in good shape.

Well, I actually prefer the bacula console in contrast to bloated and 
incomplete GUIs of most backup systems. But you are of course right in 
saying, that having one is important for a larger user base.  And for 
complicated restore-selections a GUI might come in handy too. I didn't use 
it yet, but I guess the wx-GUI can at least do this.

Regards,
Georg 

-
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


Re: [Bacula-users] Migrating to Bacula

2006-12-06 Thread Georg Altmann


--On Dienstag, 5. Dezember 2006 13:37 +0100 Rudolf Cejka 
<[EMAIL PROTECTED]> wrote:
>> With technology like LTO-3, perhaps it is time
>> to consider giving the user the ability to turn
>> off the CRC32 generation and comparison.

Agreed. LTO verifies the written data on the fly. No need for verifying or 
CRCs for bacula here. I'm not so shure about the bus inbetween though... 
What if data gets corrupted say on the SCSI bus/controller/whatever. Is 
consistency guaranteed by the SCSI protocol?
And what about the fd<->sd path? Is there an additional checksum protecting 
the data going over the network (despite of TCP)?

> It would be very good thing. I tried to patch Bacula personally,
> but then I was afraid to use it in my real environment, until it is
> tested and until I'm sure I can restore the data too :o) The patches
> were not finished and maybe they are lost. I decided that I'm not
> sufficiently ready to do such a low level thing. I have read about
> LTO3, that "the tapes contain a strong error correction algorithm that
> makes data recovery possible when lost data is within one track or up
> to 32 mm of the tape medium." Another interesting thing is that LTO3
> tape drive uses flying tape over the heads (like in hard drives) as much
> as possible, so cleaning tapes are almost not used in effect. HP says
> that the drive can run up to 6 000 hours without cleaning tape use (where
> just small internal brush is sufficient), which was big surprise for me.
> Unfortunatelly I have got this information after I have bought 5 cleaning
> tapes :o) Imagine, that one cleaning tape may be exhausted even up to
> after 90 000 hours = 3750 days >= 10 years... :o)

Actually cleaning is even harmful to the drive's heads. You should _never_ 
clean an LTO drive unless it requests a cleaning tape, as indicated by 
front LED, Tape Alert or whatever.
We use a LTO1 drive and only backup about 100 GB a week. The drive is in 
service for over one year now and never needed to be cleaned so far.

Regards,
Georg

-
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


Re: [Bacula-users] scalability / multiple storage devices / infiniBand

2006-12-07 Thread Georg Altmann


--On Donnerstag, 7. Dezember 2006 8:17 Uhr -0500 John Drescher 
<[EMAIL PROTECTED]> wrote:

> On 12/7/06, Sebastian Hegewald <[EMAIL PROTECTED]> wrote:
>> hey,
>>
>> i have 3 questions.
>>
>> 1. delivers the FD the backupdata straight to the SD or uses he the
>> detour over the Director?
> The fd sends data to the sd.

Which has been explained in multiple posts recently btw...

>>
>> 2. is it possible to write a backup of one client simultaneously to
>> different Storages Devices, so that the backup is distributed?
>>
> Yes.
Er, depends on what yo understand by "simultaneously/distributed". If you 
mean several copies on different backup media, the migration/copy feature 
of bacula is probably what you want. But AFAIK only the migration feature 
is implemented in the current development code. I.e. the migrated job will 
be on both volumes, but the original job record is purged from bacula's 
database. The volume with the original will therefore be recycled by 
following jobs, if all jobs on the volume are purged, unless you mark it 
read-only manually. If you do that, you still lost the information which 
original job is on which media. Probably not very useful in a realistic 
scenario...
So I guess you have to wait until real copy jobs are implemented. This is 
something I would love to have too... :-)
The workaround might be running two full jobs to different pools/devices, 
if one can afford the overhaed of running the same full job twice. 
Differential/incremental jobs would of course be related to the last full 
job run and not to any of the two, as I would expect it to be the case for 
copied jobs.
You cannot write a job multiplexed to multiple devices in parallel, if that 
is what you mean.


>> 3. is it possible to run bacula over a infiniBand network?
>>
bacula doesn't care for the hardware layer. Its just plain TCP/IP. If 
infiniband supports that, I see no problem here.


Regards,
Georg 

-
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


Re: [Bacula-users] Could not connect to Storage daemon on bacula:9103. ERR=Operation not permitted

2006-12-07 Thread Georg Altmann


--On Donnerstag, 7. Dezember 2006 15:28 +0100 Jon Ingason 
<[EMAIL PROTECTED]> wrote:

> I have installed new bacula server 1.38.11 on debian testing/unstable
> runnig on PC and one of the clients, running on FreeBSD 6.1 version
> 1.38.5_1, is giving me problem:
>
> run job="client-100" fileset="client File Set" level=Incremental
> client="client-100-fd" pool="client-100" when="2006-12-07 12:59:47"
> where="" storage="Fileclient-100"Run Backup job
> JobName:  client-100
> FileSet:  Client File Set
> Level:Incremental
> Client:   client-100-fd
> Storage:  Fileclient-100
> Pool: client-100
> When: 2006-12-07 12:59:47
> Priority: 14
> OK to run? (yes/mod/no):
> Job started. JobId=51
> 07-Dec 12:59 bacula-dir: No prior Full backup Job record found.
> 07-Dec 12:59 bacula-dir: No prior or suitable Full backup found. Doing
> FULL backup.
> 07-Dec 12:59 bacula-dir: Start Backup JobId 51,
> Job=client-100.2006-12-07_12.59.50
> 07-Dec 12:58 client-100-fd: client-100.2006-12-07_12.59.50 Warning:
> bnet.c:853 Could not connect to Storage daemon on bacula:9103.
> ERR=Operation not permitted
> Retrying ...

> I have have other clients which are working well. It has probable
> nothing to do with the client, but with the storage daemon. I have gone
> through the configurations files and so on, but can't find anything wrong.
>
> Are there someone who know more and can explain what can be wrong?

The bacula fd cannot connect to the storage daemon. That's also exactly 
what the log says.
Make sure any firewall between the file daemon and the storage daemon 
allows the file daemon to connect to the storage daemon on port 9103.

Regards,
Georg

-
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


Re: [Bacula-users] Copying volumes

2006-12-08 Thread Georg Altmann


--On Donnerstag, 7. Dezember 2006 14:58 -0500 Andrew Fabian 
<[EMAIL PROTECTED]> wrote:

> Any thoughts on this scenario?  I'm not sure that a quick answer exists.
>
> I'd like to have bacula back up our collection of hosts to it's
> file-based pool, so that we can browse backups "online" and restore
> individual files quickly.
>
> I'd also like to have a complete off-site copy of the pool for disaster
> prevention.
>
> So what I think I'd like to do is: backup to the file pool, then copy
> new files to dvd's, but in a way that bacula is aware of both copies.
> Then when devices get purged from the file pool, bacula will still be
> aware of the copy on dvd.
>
> Migrating doesn't quite work because that removes new data from the file
> pool, which limits our ability to browse and restore backups
> automatically.  Manually copying files a la bimagemgr doesn't work
> because bacula is not aware of the copies, and then when the file pool
> is purged, it will think that the backups are gone for good.

Read my recent post regarding the subject "scalability / multiple storage 
devices / infiniBand".
Until copy jobs are implemented I can't think of a workable solution.

Regards,
Georg

-
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


[Bacula-users] Off-topic: FAT32 disaster recovery

2006-12-08 Thread Georg Altmann
Hello,

I know this is off topic, but I hope to reach some people on this list, who 
might be able to help me with the daunting task I am facing. Please excuse 
the impertinence to post it here. The whole story is a bit lengthy - just 
ignore the post in case you're not interested.

A friend of mine, who is an architectural student, contacted me because of 
a problem with his external hard drive, containing a FAT32 partition which 
he had attached to his PowerBook. Possibly because of a failure of the USB 
electronics of the hard drive, the disk was disconnected from the system 
without being properly unmounted.  He better had used bacula to backup his 
data, because apparently the file system was damaged without Mac OS X 
noticing this. Instead he ignored the error and continued to write data to 
the disk.  Later he noticed, that he could not open certain PDF files 
because they are corrupted in some way (at least Acrobat says that).

Now (to late) he realised his misery and now I have the pleasure to try to 
restore his data. The first thing to do, is of course to create an image of 
the whole partition, to prevent the destruction of further data. 
Unfortunately nice people from Redmond can get in your way here. I attached 
the FireWire disk to my desktop, booted Windows XP and in it's ingenuity, 
it detected the file system on the disk was corrupted, and ran chkdsk over 
it before I would have hit the "any key". chkdsk moved all lost clusters to 
a directory FOUND.000 containing enumerated files FILE.CHK. 
Additionally there are files with the names of directories which existed on 
the fs. Using the UNIX file utility and some shell scripting, I was able to 
sort these files into directories on an other disk, each containing one 
file-type. It appears like PDF and Photoshop files are mostly okay, but for 
example most PDF and TIFF files are corrupted.

Now, I know that there is a plethora of file system recovery tools claiming 
to be able to restore all files.  Though, I would like to know, if there is 
any chance to restore the original FAT32 file system structures after the 
disastrous chkdsk run or if I have to live with the .chk files. My 
knowledge of FAT32's internal structures is next to zero and so I am not 
sure how to proceed. Enchantingly I couldn't find any info on what exactly 
chkdsk is doing in the MS knowledge base. So I would appreciate if anyone 
could give me some information on this and to know if I have any better 
options to restore data, other than the .chk files.

My personal judgment is, that files already got corrupted by Mac OS X 
writing to the damaged filesystem and thereby overwriting blocks (clusters? 
"chains"?  whatever) it mistakenly deemed free, but in reality belonged to 
already existing files.


Anyway, this teaches us once more, that you should
a) make regular backups (doh!)
b) not rely on external hard disks for this
c) not at all use crappy filesystems such as FAT32



Thank you for reading all this!

Regards,
Georg



-
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


Re: [Bacula-users] bscan problem

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 12:26 +0100 Angela Gavazzi 
<[EMAIL PROTECTED]> wrote:

> Hi,
>
> After a HD failure and following new install, I have to rebuild some
> catalogs  from tapes in a separate new db. (the new one is utf-8)
> As I understood I can use bscan for this.
>
> Using this command:
>
> bscan -V FULL-20060901-2 -v -s -m -c /etc/bacula/bacula-sd.conf -n
> bacula-oldcoded -u bacula /dev/st0
>
> I get this result:
>
> bscan -V FULL-20060901-2 -v -s -m -c /etc/bacula/bacula-sd.conf -n
> bacula-oldcoded -u bacula  /dev/st0
> bscan: butil.c:269 Using device: "/dev/st0" for reading.
> 08-Dez 10:58 bscan: 3301 Issuing autochanger "loaded drive 0" command.
> 08-Dez 10:58 bscan: 3302 Autochanger "loaded drive 0", result is Slot 2.
> 08-Dez 10:58 bscan: Ready to read from volume "FULL-20060901-2" on
> device "DLT-8000" (/dev/st0).
> bscan: acquire.c:200 jcr->dcr=0x80b7b48
> bscan: bscan.c:281 Using Database: bacula-oldcoded, User: bacula
> ***
> bscan: bscan.c:406 Volume is prelabeled. This tape cannot be scanned.
> ***
> Records added or updated in the catalog:
>   0 Media
>   0 Pool
>   0 Job
>   0 File
>
>
> How can I rebuild a catalog from a labelled tape, which, for my
> understanding,  every tape on which bacula wrote is?
> I read the manual chapter for bscan for- and backwards many time, but i
> don't  see the error.
>
AFAIK this indicates a blank tape which only had a label written to it. I 
fear there is no data on this tape. sorry...


Regards,
Georg



-
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


Re: [Bacula-users] Migrating to Bacula

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 11:55 + Alan Brown <[EMAIL PROTECTED]> 
wrote:

> On Wed, 6 Dec 2006, Georg Altmann wrote:
>
>> We use a LTO1 drive and only backup about 100 GB a week. The drive is in
>> service for over one year now and never needed to be cleaned so far.
>
> Our LTO drives have only ever required cleaning after an incident in the
> server room (builders being careless) raised a lot of dust which was
> sucked into the racks.
>
builders in the server room? Doesn't sound good to me... ;-)

Regards,
Georg


-
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


Re: [Bacula-users] Off-topic: FAT32 disaster recovery

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 14:16 + Alan Brown <[EMAIL PROTECTED]> 
wrote:

> On Fri, 8 Dec 2006, Georg Altmann wrote:
>
>> Unfortunately nice people from Redmond can get in your way here.
>
> understatement
:-)

>> I
>> attached the FireWire disk to my desktop, booted Windows XP
>
> Now you know why you should NEVER use MS products and operating systems
> to recover corrupted MS filesystems.
I know, I tried linux and FreeBSD before. But linux somehow didn't like my 
firewire-card and FreeBSD failed to mount the filesystem, probably due to 
it's size. 200 GB FAT32, what a mess.

Regards,
Georg



-
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


Re: [Bacula-users] Off-topic: FAT32 disaster recovery

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 06:50 -0800 Robert Nelson 
<[EMAIL PROTECTED]> wrote:

> Yep Microsoft is so obviously at fault here.  First of all they designed
> FAT in the late 70s.  Shame on them for using a filesystem design similar
> to all the other systems at the time.  Then to top it off they have the
> audacity to try and repair it when the user boots the system with a drive
> attached which is corrupted.
>
> UNIX is such a better system.  Let's see what does it do in a similar
> situation?  Well fsck notices the drive needs repair, finds all these file
> fragments that are listed as allocated but not attached to any directory.
> Gee it makes up names for them and puts them in the lost+found directory.
>
> Gee seems like exactly the same thing to me.

Ok, this wasn't about starting a flame-war on Windows vs. 
Mac/UNIX/whatever, FAT vs. ext2/ufs/whatever, pointing fingers or anything. 
So please calm down.
Anyway, linux/freebsd doesn't mount a filesystem until you explicitly tell 
it to do so and it only runs fsck on filesystems in /etc/fstab. FreeBSD 
even stops the fsck run if it finds something serious and waits for manual 
intervention. Mordern filesystems (ntfs, ufs, ext2fs,...) are much more 
stable. And many of them have been available at the beginning of the 
nineties. MS in contrast happily continued to use FAT in all its OSes 
despite of its known problems. And now you still find disks in the range of 
multiple hundert GBs using a filesystem which wasn't designed for disks of 
this size. Problem is you almost have no choice, because it is in fact the 
only fs that is support on all platforms. Linux is coming up with some 
experimental NTFS support, though I wouldn't want to use that for critical 
data yet...

>
> The only real lessons to be learned here is.  If your drive has a problem
> don't continue to use it without fixing the problem.

Explain that to the casual computer user...

> Don't use an OS that
> caches data on removable drives (Windows doesn't MAC OS apparently does).
> Don't use an OS that mounts a drive that hasn't been closed properly
> (another MAC deficiency versus Windows).  Finally don't give your drive to
> someone that doesn't know what they're talking about to have it fixed.

Finally please stop making personal offenses. Thank you.

Regards,
Georg

-
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


Re: [Bacula-users] Off-topic: FAT32 disaster recovery

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 08:08 -0800 Robert Nelson 
<[EMAIL PROTECTED]> wrote:

> Yet more disinformation, Microsoft hasn't released a desktop OS in more
> than 7 years that required FAT or even selected it as the default on new
> installations.

Wow, great more than seven years, you mean like 1999? I have no idea when 
Win Me was released exactly, but for sure it didn't support anything but 
FAT. NTFS and other reliable filesystems have been around for quite a while 
back then. This timebomb of a filesystem is still ticking on a lot of 
computers.

> Yes all Microsoft Operating Systems still support FAT as do all other
> Operating Systems.  Otherwise it would be very hard to get your photos off
> the compact flash cards, etc.

Yes, even worse. Just as I sad this archaic fs is still being used. And I 
can hear people whining because they lost their data on CF, usb-thumbs and 
the like because FAT32 is used for them by default. If you pull a ufs2 
formatted drive which is being written to for example, you might loose some 
files, but it is very unlikely to loose the entire tree, in contrast to 
FAT. With soft-updates enabled you probably won't loose anything at all, 
except the inodes which have actually been written to. I'm not sure about 
NTFS as I have never used it on removable drives, but I guess it is able to 
handle an unexpected unmount gracefully too.

> As for your proverbial "casual user", let's use a car as an example.
>
> Someone is driving down the street and the engine starts making a funny
> noise.  Does he take it to a mechanic; no he continues to drive it.  Now
> the warning lights are starting to come on in the dash, is it time to
> seek help? Nope he keeps driving.  Finally the car stops running, must be
> that d*mn Ford engine.
>
> Finally, regarding my other comment to which you seemed to have taken
> By your own admission you've said you don't understand the internals of
> filesystems, you don't know what chkdsk does, and you don't know about the
> various tools available for fixing corrupted drives.

Oh yes, I do know quite a bit about filesystems. But I don't usually bother 
myself with FAT and blame other people for not doing so.
Therefore I have never been in the situation of having to resort to 
esotheric recovery utilities, offered by companies profiting from users, 
who don't know to what tipsy filesystems they entrust their data to.
And for your car example: A car inspection doesn't usually cost thousands 
of dollars. A data recovery by a professional company will. So I guess it's 
wise to turn to a friend who knows to handle the situation instead of 
tampering around. Admitted booting windows with the drive attached was a 
bit naive, but I didn't intend to shredder the disk with chkdsk. Merely I 
was trying to create an image to conserve the data on the disk. 
Unfortunately, as I explained, windows knew better.

Now please stop bugging me with your "I'm the greatest geek in town"-gossip 
and go play flame-war somewere else.

Thanks to the others for their constructive comments.

Regards,
Georg

-
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


Re: [Bacula-users] Backing up with two Bacula servers?

2006-12-08 Thread Georg Altmann


--On Freitag, 8. Dezember 2006 08:59 -0800 Willard Farqwark 
<[EMAIL PROTECTED]> wrote:

> I have been running Bacula writing to disk for many months now.  Very
> fast and easy fast restore. I am building a new server with Bacula
> writing to a LTO-2.
> Question is - How do I set up a client Bacula-fd for two servers?

I suspect you mean two storage daemons. You don't. According to the fine 
manual

you just add a new storage resource in your director configuration pointing 
to the storage deamon with the LTO2 device. Now you can use that storage 
definition in your job/schedule definitions. Of course you need to have the 
second sd configured and running for this to work.

Regards,
Georg

-
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


Re: [Bacula-users] Migrating to Bacula

2006-12-09 Thread Georg Altmann


--On Samstag, 9. Dezember 2006 00:51 +0100 Arno Lehmann <[EMAIL PROTECTED]> 
wrote:

> Hi,
>
> On 12/5/2006 12:04 PM, Kern Sibbald wrote:
>> On Tuesday 05 December 2006 11:04, Rudolf Cejka wrote:
>>
>>> Arno Lehmann wrote (2006/12/01):
>>>
 On 11/30/2006 9:24 PM, Per Andreas Buer wrote:
> ...I would be very interested to know why you don't like Areika.  Having
>> this information always helps ...
>
> Generally, I found Arkeia to be unreliable. There were simply too many
> occasions where backups didn't run, too often without any clues in the
> log files.
>
> The license system especially caused all sorts of trouble, like it
> insisted on not having licenses for machines it had in its database,
> only these machines weren't there. Or changing the machine type without
> aparent reason, and thus requiring a more expensive license.
>
> The GUI looked interesting, but I couldn't work with it.
>
> The tools to manage Arkeia from the shell use a rather peculiar syntax,
> at least very different from any unix/gnu software I know.
>
> The catalog is held in disk directories as links, files, and
> directories. This is a performance problem for the relatively
> underpowered backup servers I run. Also, it's not easily separated
> between machines (which is a feature I really like in Bacula, because I
> have to lose two servers before I need a bare-metal Bacula server
> restore), and managing it when the provided tools don't work correctly
> is impossible.
>
> Some of the components tended to crash, which lead to all sorts of
> problems hidden behind a GUI, rather than presented by it.
>
> The GUI was really painful - setting up a basic configuration needed
> many kilometers of mouse abuse, and I never found a good overview of
> what I set up. This might be a matter of taste, but I do like text
> consoles for system management tasks.
>
> Support was not really helpful - very often it was advised to stop
> Arkeia, remove some things from the file system, and restart. Or use the
> command line tools to manipulate the catalog (usually deleting hosts or
> changing licenses).
>
> The licensing scheme is simply too expensive for me and my customers.
> After all, they need their money to pay me :-) and paying for Arkeia
> software does not deliver you a Backup solution that works out of the
> box, so they need support anyway.
>
> Much of the above is also reflected on the arkeia users mailinglist IMO,
> so I assume that it's not only my fault that I couldn't work with Arkeia.

Your descriptions sound familiar. I was running Netvault on FreeBSD before 
bacula.
I remember one bug where netvault would actually count the virtual library 
capacity (i.e. disk based storage) twice. This lead to the license not 
being sufficient and I had to beg support for a extended license to 
workaround the bug. They did never fix this bug in over 3 years.

Generally netvault support seemed very disorganized.
After a hardware change netvault wouldn't start up at all anymore but 
segfault while scanning the scsi busses. I couldn't get this fixed after a 
lot of struggling with support and in the end I finally decided to dump 
netvault alltogether for something else. It seemed like FreeBSD wasn't the 
best supported platform for netvault after all.

Enough ranting on netvault. I'm very happy with bacula. There has not been 
a single problem since the setting it up November 2005. Still running 
bacula 1.38.2 btw.

I would be interested in other experiences with netvault if anybody else 
has used it before. Maybe it's a better experience on other platforms or 
hardware?

Regards,
Georg

-
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


Re: [Bacula-users] (no subject)

2006-12-09 Thread Georg Altmann


--On Samstag, 9. Dezember 2006 10:18 +0100 Guy Corbaz <[EMAIL PROTECTED]> 
wrote:

> Hello,
>
> I'm using Bacula since several days on Ubunto 6.06. Suddenly, after
> everything worked fine, I'm getting the following messages:
>
> 09-Dec 01:05 saturne-dir: Pbx.2006-12-09_01.05.00 Fatal error:
> sql_create.c:91 sql_create.c:91 insert INSERT INTO Job
> (Job,Name,Type,Level,JobStatus,SchedTime,JobTDate) VALUES
> ('Pbx.2006-12-09_01.05.00','Pbx','B','I','C','2006-12-09
> 01:05:00',1165622700) failed:
> MySQL server has gone away
> 09-Dec 01:05 saturne-dir: Pbx.2006-12-09_01.05.00 Fatal error:
> sql_create.c:93 Create DB Job record INSERT INTO Job
> (Job,Name,Type,Level,JobStatus,SchedTime,JobTDate) VALUES
> ('Pbx.2006-12-09_01.05.00','Pbx','B','I','C','2006-12-09
> 01:05:00',1165622700) failed. ERR=MySQL server has gone away
>
> If I restart the server and re-launch manually the backup, it works, but
> crash again during the night. Manually restarting MySQL is not enough.

Please supply version information of bacula and mysql if asking questions. 
Also your configuration files of bacula-dir and mysql might be useful.

Now to the problem: The log looks like you are running a job and bacula 
fails to insert the job into the databse because the connection dropped or 
mysqld died.
This problem repeatetly been reported for mysql version 5. If you run mysql 
version 5 try downgrading to version 4.
What do you mean by "restart the server"? Reboot the machine or restart the 
deamon process(es) of mysql/bacula? Please give more detauled information 
here.
What does the mysqld log say?
Try running mysqld with debugging enabled to find out why it dies (if it 
does actually) or why the connection is dropped.
Do you run other applications using the mysql server?
Is the mysql server on the same machine? If not, this might boil down to 
networking problems.

And BTW, google is your friend:



This is what I can think of currently. Any other ideas?

Regards,
Georg





-
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


Re: [Bacula-users] high availability

2006-12-11 Thread Georg Altmann


--On Montag, 11. Dezember 2006 11:50 +0100 Sebastian Hegewald 
<[EMAIL PROTECTED]> wrote:

> hello,
>
> i should compare bacula and amanda concerning high availability.
>
> now my question is. is there support for a failover-scenario?
> just a little example:
>
> i have a aktiv director and some clients and storagedaemons. now the
> director fails. is it possible to switch to a passiv director so that he
> will be the aktiv one.

Unless you are thinking of "real-time" failover, i.e. during running jobs, 
I think this is pretty straightforward:
Basically all stateful information of bacula is stored in the database. If 
you keep a stand-by director on a second machine with the configuration 
synced (mirrored storage for example, depends on how far you want to 
go...), all you need to do is to switch the IP. I'm not really familiar 
with failover-technology, but there sould be solutions available for 
switching the IP over to the standby machine.

But actually I don't see the reasoning behind this. The director really is 
the easiest part of bacula to replace (config-files+ip) and probably the 
least likely to fail. You might rather have a database with failover 
support and a sd+tapedrive on standby. AFAIK MySQL can do this, not sure 
about PostgreSQL.

Regards,
Georg

-
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


Re: [Bacula-users] problems with hardlinks to "schg" secured files on FreeBSD on restore

2006-12-12 Thread Georg Altmann


--On Dienstag, 12. Dezember 2006 06:26 +0100 Oliver Lehmann 
<[EMAIL PROTECTED]> wrote:

> Hi,
>
> I've noticed that when I'm backing up files with the "schg" flag set, and
> also files which are a hardlink to such a file with a "schg" flag the
> restore does not work.
> It looks like bacula first restores the file with the "schg" flag, and
> then tries to recreate the hardlink which failes because of the "schg"
> flag.
>
> [EMAIL PROTECTED] bacula> ls -lo /mnt/files/.bacula/restores/gurke/usr/bin
> [...]
> -r-xr-xr-x   1 root  wheel   -   9516 May 12  2006 chkey*
> -r-sr-xr-x   1 root  wheel   schg   18140 May 12  2006 chpass*
> -r-xr-xr-x   1 root  wheel   -  72220 May 12  2006 ci*
> [...]
> -r-xr-xr-x   1 root  wheel   -   2006 May 12  2006 pagesize*
> -r-sr-xr-x   1 root  wheel   schg5920 May 12  2006 passwd*
> -r-xr-xr-x   1 root  wheel   -   5792 May 12  2006 paste*
> [...]
>
>schg, schange, simmutable
>set the system immutable flag (super-user only)
>
>
> The error messages are:
>
> 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58 Error:
> create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/yppasswd ->
> /mnt/files/.bacula/restores/gurke/usr/bin/passwd: ERR=Operation not
> permitted 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58
> Error: create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/ypchsh ->
> /mnt/files/.bacula/restores/gurke/usr/bin/chpass: ERR=Operation not
> permitted 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58
> Error: create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/ypchfn ->
> /mnt/files/.bacula/restores/gurke/usr/bin/chpass: ERR=Operation not
> permitted 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58
> Error: create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/ypchpass ->
> /mnt/files/.bacula/restores/gurke/usr/bin/chpass: ERR=Operation not
> permitted 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58
> Error: create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/chsh ->
> /mnt/files/.bacula/restores/gurke/usr/bin/chpass: ERR=Operation not
> permitted 10-Dec 20:31 nudel-fd: client-nudel-files-r.2006-12-10_20.28.58
> Error: create_file.c:308 Could not hard link
> /mnt/files/.bacula/restores/gurke/usr/bin/chfn ->
> /mnt/files/.bacula/restores/gurke/usr/bin/chpass: ERR=Operation not
> permitted
>
> I'm running bacula 1.38.11 on FreeBSD 6.1-STABLE.
> Do you have any Idea how to fix this? If you need the configuration
> please tell me, but I don't think so since it looks to me like an
> config-independent, general, problem.

Same error using bacula-1.38.2 on FreeBSD 4.11 here.

Just to clarify the problem - looking at the bacula code it seems to me 
like the follwing happens:
- fd extracts file A
- d sets the attributes/flags for A
- d restores file B which is a hardlink to A
The last operation fals because A has the SF_IMMUTABLE (system 
immutable/schg) flag set.

I'm speculating this will occur on all UNIX/linux/BSD platforms which have 
an IMMUTABLE attribute/flag for files. Can somebody confirm this?

Looks like bsdtar had this problem too:


And even (Net?)BSDs restore:


The solution proposed is to keep file flags in a list and set them after 
all files have been restored, so that all hardlinks can be created. Though 
it is probably enough to do this for hardlinked files. Looks like this 
would make the restore code even more complicated.

But what is the solution if you restore a file from an incremental backup 
which is a hardlink to an IMMUTABLE file?
bacula could clear the IMMUTABLE flag for creating the hardlink and re-set 
it afterwards. But this is not possible if kern.securelevel > 0. I guess it 
would be wrong to do so anyway, because it undermines the security of the 
system. After all files are set IMMUTABLE for a reason. Consider the fd 
crashes for some reason after it cleared the flag... is there an atomic way 
to do this?

My solution for restoring a freebsd system right now is to reinstall the 
base system after the restore (make buildworld installworld).

Regards,
Georg

-
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


Re: [Bacula-users] problems with hardlinks to "schg" secured files on FreeBSD on restore

2006-12-12 Thread Georg Altmann


--On Dienstag, 12. Dezember 2006 19:40 +0100 Oliver Lehmann 
<[EMAIL PROTECTED]> wrote:

> Kern Sibbald wrote:
>
>> Not only is it more complicated, but it can be very expensive.  Suppose
>> you  are restoring a  million hard linked files.  That could mean that
>> you need  you will need to keep a list that could amount to hundreds of
>> millions of  bytes (you need to keep at least forward links, the full
>> path and the file,  as well as certain of the file attributes).   If you
>> are restoring 5 or 10  million files, it could even be worse.  In
>> addition, if you have multiple  files linked together (i.e. more than
>> 2), it would add more complication  because when hitting the second
>> linked file in the list, the real file would  already be marked
>> immutable.  You can solve that by having a doubly linked  list in binary
>> order, but then the computation costs go way up.
>
> Why not just keep a list of filenames which should have the IMMUTABLE
> flag, and apply this flag after the whole backup is done to those files.

If you read my post carefully, that is what I (or rather the NetBSD 
posting) proposed... ;-)
As far as I understand the problem, all you need, is a dictionary/tree of 
paths for storing the attributes. Then you can set the attributes after the 
restore is finished.
I'm not saying this is easy to implement, though.

>> Off hand, I would say that the FreeBSD guys have worked themselves into
>> a big  hole.   They have violated the rule that says that root can do
>> anything.

Using linux with MAC (mandatory access control), or one of the other dozen 
of similar implementations, basically does the same thing. The FreeBSD 
problem is probably just one of many which will arise with all those policy 
frameworks. It appears like nobody thought of the implications for backups 
and especially restores this causes. Maybe a command to mark a directory 
tree "restorable", i.e. so that all policies have no effect, would be nice. 
But that's just a wild guess...


Regards,
Georg


-
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


Re: [Bacula-users] problems with hardlinks to "schg" secured files on FreeBSD on restore

2006-12-12 Thread Georg Altmann


--On Dienstag, 12. Dezember 2006 19:29 +0100 Oliver Lehmann 
<[EMAIL PROTECTED]> wrote:

> Georg Altmann wrote:
>
>> The solution proposed is to keep file flags in a list and set them after
>> all files have been restored, so that all hardlinks can be created.
>> Though  it is probably enough to do this for hardlinked files. Looks
>> like this  would make the restore code even more complicated.
>>
>> But what is the solution if you restore a file from an incremental
>> backup  which is a hardlink to an IMMUTABLE file?
>
> Or what is when you just try to restore the hardlink (and you didn't knew
> it was a hardlink). Beside saying that reinstalling the OS should be
> recommended here, the IMMUTABLE flag can also be used for non-OS files
> too. This can't be "fixed" by just reinstalling the OS ;)

No, of course not. I was just supposing this as a fix for your specific 
error messages.
On the other hand, having files with the schg flag set AND hardlinks is not 
something that happens very often.

> I'd say that
> keeping the flags and setting them when the whole restore is done is one
> step in the right direction and fixes at least some of the problems. Even
> if it does not fix the problems 100%

Just read my other post about SELinux and MAC. I don't think managing 
immutable flags is the responsibility of bacula or any other backup 
software. There should really be a possibility to make backups and 
restores, without having all those policies to get in your way, instead of 
implementing a complicated handling scheme in every backup-software. Maybe 
worth a bug report for FreeBSD?

Cheers,
Georg

-
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


Re: [Bacula-users] problems with hardlinks to "schg" secured files on FreeBSD on restore

2006-12-12 Thread Georg Altmann


--On Dienstag, 12. Dezember 2006 21:24 +0100 Oliver Lehmann 
<[EMAIL PROTECTED]> wrote:

> Georg Altmann wrote:
>
>> There should really be a possibility to make backups and
>> restores, without having all those policies to get in your way, instead
>> of  implementing a complicated handling scheme in every backup-software.
>> Maybe  worth a bug report for FreeBSD?
>
> But that would "bypass" the goal those flags are trying to reach,
> wouldn't it?

No, not really. You can clear the schg flag as root as you said your self. 
Why shouldn't it be possible for root to "disable" immutable flags for a 
whole tree?
Obviously for securelevel > 0 this should be prohibited. And for MAC you 
could have just another policy (JAO) for allowing this.
I'm not a kernel hacker, so I don't know if this can be implemented in some 
reasonable manner - it's just an idea...

Regards,
Georg

-
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


Re: [Bacula-users] (no subject)

2006-12-14 Thread Georg Altmann


--On Mittwoch, 13. Dezember 2006 15:44 -0700 "Sprague, Scott" 
<[EMAIL PROTECTED]> wrote:

[--SNIP--]
>
> Our backups are done on a machine named gdcvault.  This machine was
> recently upgraded from Mandriva 2006 to 2007.  After the update, the
> backups on one of our clients, orange, have failed with the error message
> below.
>
> _
> _
>
> 13-Dec 21:59 gdcvault-dir: orange_Full_Daily.2006-12-13_21.59.13 Error:
> bnet.c:257 Read error from File daemon:orange1.gdc:9102: ERR=Connection
> reset by peer
>
> 13-Dec 21:59 gdcvault-dir: orange_Full_Daily.2006-12-13_21.59.13 Error:
> Bacula 1.38.11 (28Jun06): 13-Dec-2006 21:59:21

[--SNIP--]

> There is no firewall on either machine, both machines are on the same
> network, and we had no problems with the backups on this client until the
> upgrade.  Bacula version 1.38.11 is running on the backup server; the
> client is running 1.34.5.  I would suspect this as the problem, except
> that we have multiple Linux clients running the same version and having
[--SNIP--]

You should not do that. Make sure fd version == dir version == sd version.

Please consider turning off html in your mail-client when posting to 
mailing lists.

Regards,
Georg

-
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


Re: [Bacula-users] Postgres slow because of autocommit

2006-12-14 Thread Georg Altmann


--On Mittwoch, 13. Dezember 2006 15:57 +0100 Gabriele Bulfon 
<[EMAIL PROTECTED]> wrote:

>
> I came across the performance problem with SPARC Sun280R machines, at
> first, because they have a slower processor then modern AMD or x86
> machines.
> On sparc machines large backups cannot be managed, because the transfer
> rate of the LTO is slowed down up to 1Mb/s or lower, because of the slow
> performance of Postgres on the same machine.
> Moving the DB on another x86 / amd machine solve the problem, but I
> anyway can reach a maximum rate of 8-10Mb/sec, that is not the maximum
> transfer rate I may have on those LTO drives (that may run from 15 to 60
> to 120 Mb/s depending on the version).
>
> On x86/amd machines, the transfer is higher and acceptable, but yet not
> using the full potential of the scsi library or tape drive, that is
> anyway lower than it may be, always because postgres is slowing down the
> bacula process.

Have you made sure it's not a configuration or hardware problem? First off, 
I have not expirience with SPARCs. Still, you might check the following:

Did you test your LTO device with dd? Just create a file with say 1GB of 
random data and dd it to the tape (with no other load). Blocksizes of 32k 
or 64k should do. For LTO1 this should give you something like 15 MB/s - 
regardless of compression settings of the device. If not, either your disks 
are slow or I suspect you have hardware trouble. Looking at the machine's 
specs, I think it should easily handle this data rate.

Make sure the machine is not swapping. If it does, buy some RAM... :-)

Have you tested postgres performance with something different than bacula?

Regards,
Georg





-
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


Re: [Bacula-users] (no subject)

2006-12-14 Thread Georg Altmann


--On Donnerstag, 14. Dezember 2006 11:42 +0100 Georg Altmann 
<[EMAIL PROTECTED]> wrote:
> --On Mittwoch, 13. Dezember 2006 15:44 -0700 "Sprague, Scott"
> <[EMAIL PROTECTED]> wrote:
> Please consider turning off html in your mail-client when posting to
> mailing lists.
Oh, and a descriptive subject line would be nice too. Thanks.

Regards,
Georg

-
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


Re: [Bacula-users] Slow positioning at the end of the tape?

2006-12-16 Thread Georg Altmann


--On Friday, December 15, 2006 14:07:01 +0100 Jose Molina 
<[EMAIL PROTECTED]> wrote:

> El vie, 15-12-2006 a las 11:48 +, Alan Brown escribió:
>> On Fri, 15 Dec 2006, Jose Molina wrote:
>>
>> > 14-Dec 23:02 rap-sd: Volume "Diario-02" previously written, moving to
>> > end of data.
>> > 14-Dec 23:46 rap-sd: Ready to append to end of Volume "Diario-02" at
>> > file=48.
>>
>> This is wildly wrong with LTO of any type.
>>
>> > Are these numbers normal?
>>
>> No.
>>
>> The LTOs here take less than 3 minutes to position to any point on the
>> tape

same here

> OMG... ok, these are the configs i've done. Also i've always used btape
> to test my configs, and btape said everything was Ok (drive and
> library):
>
> BTW, it's LTO 2...
>
> Tapeinfo, (got variable block size ok, etc..)
> [EMAIL PROTECTED] /usr/local/etc]# tapeinfo -f /dev/pass0
> Product Type: Tape Drive
> Vendor ID: 'HP  '
> Product ID: 'Ultrium 2-SCSI  '
> Revision: 'S54W'
> Attached Changer: No
> SerialNumber: 'HU106371UC'
> TapeAlert[50]:Undefined.
> MinBlock:1
> MaxBlock:16777215
> Ready: yes
> BufferedMode: yes
> Medium Type: Not Loaded
> Density Code: 0x42
> BlockSize: 0
> DataCompEnabled: yes
> DataCompCapable: yes
> DataDeCompEnabled: yes
> CompType: 0x1
> DeCompType: 0x1
> BOP: yes
> Block Position: 0
>
> This is my drive config that works on BSD (trickier than configuring on
> linux...i'm not really sure i got all the correct directives on it...)

I'm running LTO1 (HP) von FreeBSD 4.11 wihout any trouble here.
My device config looks a bit different then your's, though I cannot say 
wether your's are really wrong.
Settings marked with (*) are just missing in my config, have a look at 
these.

># Drive MSL (HP Ultrium LTO 2)
> Device {
>   Name = Drive-1  #
>   Drive Index = 0
>   Media Type = LTO2
>   Archive Device = /dev/nsa0
>   AutomaticMount = yes;   # when device opened, read it
>   AlwaysOpen = yes;
>   RemovableMedia = yes;
>   RandomAccess = no;
>   AutoChanger = yes
>
>   Offline On Unmount = no
>   Hardware End of Medium = no
>   (*)BSF at EOM = yes
>   (*)Backward Space Record = no
>   (*)Fast Forward Space File = no
>   (*)TWO EOF = yes
> }

Did you add
/usr/bin/mt -f /dev/nsa0 seteotmodel 1
to /etc/rc.local? That was necessary for our drive, but your btape tests 
should have failed with a wrong setting, so this might not be it.

Try seeking using mt.

Regards,
Georg


--

Georg Altmann <[EMAIL PROTECTED]>
LAS-CAD GmbH, Munich, Germany
http://www.las-cad.com


-
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


Re: [Bacula-users] bacula-sd config for tape

2007-01-22 Thread Georg Altmann


--On Montag, 22. Januar 2007 12:15 -0600 Zeratul <[EMAIL PROTECTED]> wrote:

>>
>> Seems mtx-changer related problem?
>> Did you do the autochanger tests with btape?
>> Can you post SD config?
>>
>
> It's not an mtx problem. The autochanger is working well - I can use mtx
> command to get the status of the changer, to load/unload tapes, to move
> tapes from one slot to another. More than this, the local bacula-dir is
> working also well with the autochanger. It labeled the tapes using "label
> barcode", it's  saving all backups successfully on the tapes and I can
> restore everything.

The error was:
3908 Error scanning autocharger list command: autochanger slots 
Ultrium-Drive"

Check the list command in your mtx-_script_. See 


Regards
Georg

-
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


Re: [Bacula-users] DVD+RW vs DVD-RAM for incremental daily backups

2007-01-27 Thread Georg Altmann


--On Freitag, 26. Januar 2007 20:27 -0500 cy tune <[EMAIL PROTECTED]> wrote:

> On 1/26/07, Arno Lehmann <[EMAIL PROTECTED]> wrote:
>> Hello.
>
> Hi
>
>> On 1/27/2007 12:23 AM, cy tune wrote:
>> > Will DVD+R, DVD+RW, and DVD-RAM all work fine when I need multiple
>> > disks for each task?  If a weekly backup needs 3 disks, will that be
>> > okay? Similarly for the other tasks.
>>
>> You mean, if you can mix the three media types? No. You can mix DVD+R
>> and +RW but DVD-RAM is treated differently.

If I remember correctly, growisofs happily writes to a dvd-ram, as long as 
your drive supports it. Please correct me if I'm wrong.

> I didn't phrase it properly.  I meant can I backup to several disks of
> the same type for one task?  If my daily backup takes 2 DVD+R disks,
> is that okay?

I'm not familiar with bacula dvd-writing... I leave this question for the 
bacula-dvd-versed to answer. :-)

>> > Is there any different setup you would recommend?  I looked into tape
>> > drives but they are so expensive for the tapes and drives.
>>
>> The have much higher capacity, are more reliable, and more robust... I
>
> I'm not sure I would use the capacity though.
>
> It's about $8 per disk for a Verbatim double sided DVD-RAM disk (9.4
> GB) in a cartridge.  Going the tape route would cost far, far more
> wouldn't it?

For small amounts of data (say a few GB) I would say this is true. DVD 
devices are cheap. Though nothing beats tape media comparing price/GB, 
still...

>> prefer tape or, if offsite storage and the ability to endlessly add
>> storage count less than speed (and price) disk.
>
> I'm not concerned about speed of a tape vs DVD.  I am price conscious
> at this point though.  My old (unfortunate) method was a full backup
> every 6 months with no incremental backups.  I'm now looking at using
> bacula for daily backups and weekly/monthly full backups.
>
> I had a tape drive on an older computer and never used it.  It's
> capacity is so small compared to today's hard drives.  It seems to me
> that by the time I would want to replace DVD-RAM disks, I would be
> ready to upgrade my tape drive to support larger capacity tapes.  By
> the time I'm ready to replace the DVD-RAM disks, I could be buying
> blu-ray or hd-dvd or whatever is sufficiently cheap at the time.
>
> Let me know if any of this is wrong. :)  I like the bacula manual as
> far as setting everything up.  It looks clear how all the parts
> integrated and how to write a config file for each part.  What's not
> clear is what backup media to choose and what kind of backup
> strategies people use.
>
>> > with the hardware verification?  I'm not sure
>> > how to turn that on/off in Linux since it's just treated as a hard
>> > drive.  According to the wikipedia entry, it will take about twice as
>> > long to write.
>>
>> Right, it takes longer and it's always on. This is not something you can
>> turn on or off, AFAIK.

The built-in verification makes DVD-RAM the most reliable medium in the 
DVD-zoo. Though it is also the slowest. I have seen far too many unreadable 
DVD+RW discs. I wouldn't recommend these for backups. Double-sided DVD-RAMs 
are also problematic. You cannot physically label them and its difficult to 
handle them without making fingerprints. Either use a drive that handles 
caddies or use single-sided discs. The former option is of course the 
safest.

I'm using a LG DVD-RAM drive with single-sided DVD-RAMs on my private 
computer (Windows) for archiving data (not bacula). Using UDF seemed the 
best option and I'm quite happy with it. No bad discs so far.

Using DVD-RAM in combination with bacula is on my projects list. Is anybody 
actually doing this?

Regards,
Georg

-
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


Re: [Bacula-users] DVD+RW vs DVD-RAM for incremental daily backups

2007-01-30 Thread Georg Altmann


--On Montag, 29. Januar 2007 18:07 +0100 Cosimo Streppone 
<[EMAIL PROTECTED]> wrote:

>> Using DVD-RAM in combination with bacula is on my projects list. Is
>> anybody actually doing this?
>
> Yes, I am doing this on my home Linux workstation.
> I use a Samsung GSA-2164D dvdrecorder with a plain (no cartridge) DVD-RAM
> disk.
> I'm in the first stage of "try-and-see-what-happens" and I configured
> DVD-RAM device exactly like a usb removable drive, with its mount
> point `/mnt/dvdrecorder'.
> I should only make sure that disc is already mounted when I start the
> bacula job.
> But that is easily solvable with a RunBeforeJob directive (or it should
> be).

Actually I would rather like to use DVD-RAM as a raw device. Just the way 
normal DVD-writing works. What's the point of having a filesystem with one 
file on it? ;-)

Thanks for your post, though.

I have not yet explored how to use raw block devices with bacula (tape 
only). DVD-RAM are random access. So, can I simply use Archive Device = 
/dev/mydvdram and Device type = file?
I guess I just have to try it out. I'll let you know how it goes.

> I found the DVD-RAM media to be somewhat slow when writing full
> directory subtrees, but writing to a file like bacula does works
> smoothly...

This depends very much on the filesystem you use. Most filesystems store 
file meta-data in fixed places. Therefore the drive has to seek across the 
disc to update the meta-data when writing new files (or when making changes 
to the fs in general). This leads to bad performance and also makes the 
medium wear out faster at these sectors. Though I am wondering if the 
latter  one is an issue with DVD-RAMs at all, because of automatic 
relocation of bad sectors and verification. If somebody has a deeper 
insight into this, I would be pleased to be enlightened.
UDF was designed for use with (re-)writable optical media. It writes to the 
disc in a sequential manner. There is no need for the drive to seek across 
the disc and therefore you get better performance. AFAIK linux now has good 
support for UDF, so maybe you want to give it a try.
Sadly, FreeBSD still only has read support. :-(((


Regards,
Georg

-
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


Re: [Bacula-users] Compatibility Bacula ->Tandberg StorageLoader LTO2

2005-11-20 Thread Georg Altmann



--On Thursday, November 17, 2005 23:56:35 +0100 [EMAIL PROTECTED] 
wrote:



We want to work with a Tandberg StorageLoader LTO2.
It have 8 tapes and one drive.
I have looked at the list with the known compatible autoloaders
but I cant find any information about this one.

Does anyone know whether this Autoloader works with Bacula and Suse 9.3

I am happy with any Information.

Maybe someone has another tip for autoloader in that size working with
Bacula.


Maybe this helps:

http://www.bacula.org/dev-manual/Supported_Tape_Drives.html

Regards,
Georg

--

Georg Altmann <[EMAIL PROTECTED]>
LAS-CAD GmbH, Munich, Germany
http://www.las-cad.com



---
This SF.Net email is sponsored by the JBoss Inc.  Get Certified Today
Register for a JBoss Training Course.  Free Certification Exam
for All Training Attendees Through End of 2005. For more info visit:
http://ads.osdn.com/?ad_id=7628&alloc_id=16845&op=click
___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users


[Bacula-users] btape fill failure with LTO-3 on FreeBSD 9.3

2015-01-04 Thread Georg Altmann
Hey,

I am having trouble getting an HP LTO-3 tape working on FreeBSD 9.3
amd64. The normal btape test finishes successfully but the btape fill
test with two tapes fails. The test fails when btape tries to mount
TestVolume1 after having written TestVolume1 and TestVolume2.

Mount first tape. Press enter when ready:
btape: btape.c:2519-0
04-Jan 17:33 btape JobId 0: Warning: acquire.c:276 Read acquire:
vol_mgr.c:382 Could not reserve volume "TestVolume1" for append, because
it will be read.
Mount Volume "TestVolume1" on device "LTO3-0" (/dev/nsa0) and press
return when ready:
04-Jan 17:33 btape JobId 0: Error: mount.c:834 Hey! WroteVol
non-zero !
btape: mount.c:835-0 Hey! WroteVol non-zero !

The prompt

Mount Volume "TestVolume1" on device "LTO3-0" (/dev/nsa0) and press
return when ready:

just re-appears every time I hit enter. I have attached the full btape
output.

What about the error message "Error: mount.c:834 Hey! WroteVol
non-zero !" ?

This is the bacula-sd configuration of the device:

Device {
  Name = LTO3-0
  Media Type = LTO3
  Archive Device = /dev/nsa0
  AutoChanger = no;
  Spool Directory = /bspool/bacula
  AutomaticMount = yes;   # when device opened, read it
  AlwaysOpen = yes;
  RemovableMedia = yes;
  RandomAccess = no;
  Hardware End of Medium = no;
}

This comes as a bit of a surprise to me, since I successfully operated
an HP LTO-1 drive with the same configuration on FreeBSD successfully.
However, this was a previous version of FreeBSD and bacula.
Operating the drive with tar and mt (fsf) works just fine.

Might this just be a regression with btape?

Does anyone run a similar configuration (FreeBSD 9.3, LTO-3)? If you
could post your bacula-sd configuration that would be great.

FreeBSD 9.3-RELEASE-p2 amd64

% pkg info bacula\*
bacula-client-7.0.5_1
bacula-server-7.0.5_2

Regards
Georg

-- 
PGP-Key: 0x1E320E65
D150 7783 A0D1 7507 1266  C5B3 BBF1 9C42 1E32 0E65

I don't like the idea of secret agencies to analyse and archive
personal communication. GnuPG is available as open source, free as as in
freedom, as a countermeasure. I use http://www.enigmail.net/ for Mozilla
Thunderbird. If you can, please use a frontend of your choice to send me
encrypted e-mail. See http://www.gnupg.org/ for an overview.



signature.asc
Description: OpenPGP digital signature
--
Dive into the World of Parallel Programming! The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users


Re: [Bacula-users] btape fill failure with LTO-3 on FreeBSD 9.3

2015-01-11 Thread Georg Altmann
I just ran a successful backup and restore spanning two LTO3 tapes with
the Device setup below. It looks like btape is broken and does not
actually reflect the volume handling code of bacula-sd.

Regards,
Georg

Am 04.01.2015 um 23:05 schrieb Georg Altmann:
> Hey,
> 
> I am having trouble getting an HP LTO-3 tape working on FreeBSD 9.3
> amd64. The normal btape test finishes successfully but the btape fill
> test with two tapes fails. The test fails when btape tries to mount
> TestVolume1 after having written TestVolume1 and TestVolume2.
> 
> Mount first tape. Press enter when ready:
> btape: btape.c:2519-0
> 04-Jan 17:33 btape JobId 0: Warning: acquire.c:276 Read acquire:
> vol_mgr.c:382 Could not reserve volume "TestVolume1" for append, because
> it will be read.
> Mount Volume "TestVolume1" on device "LTO3-0" (/dev/nsa0) and press
> return when ready:
> 04-Jan 17:33 btape JobId 0: Error: mount.c:834 Hey! WroteVol
> non-zero !
> btape: mount.c:835-0 Hey! WroteVol non-zero !
> 
> The prompt
> 
> Mount Volume "TestVolume1" on device "LTO3-0" (/dev/nsa0) and press
> return when ready:
> 
> just re-appears every time I hit enter. I have attached the full btape
> output.
> 
> What about the error message "Error: mount.c:834 Hey! WroteVol
> non-zero !" ?
> 
> This is the bacula-sd configuration of the device:
> 
> Device {
>   Name = LTO3-0
>   Media Type = LTO3
>   Archive Device = /dev/nsa0
>   AutoChanger = no;
>   Spool Directory = /bspool/bacula
>   AutomaticMount = yes;   # when device opened, read it
>   AlwaysOpen = yes;
>   RemovableMedia = yes;
>   RandomAccess = no;
>   Hardware End of Medium = no;
> }
> 
> This comes as a bit of a surprise to me, since I successfully operated
> an HP LTO-1 drive with the same configuration on FreeBSD successfully.
> However, this was a previous version of FreeBSD and bacula.
> Operating the drive with tar and mt (fsf) works just fine.
> 
> Might this just be a regression with btape?
> 
> Does anyone run a similar configuration (FreeBSD 9.3, LTO-3)? If you
> could post your bacula-sd configuration that would be great.
> 
> FreeBSD 9.3-RELEASE-p2 amd64
> 
> % pkg info bacula\*
> bacula-client-7.0.5_1
> bacula-server-7.0.5_2
> 
> Regards
> Georg

-- 
PGP-Key: 0x1E320E65
D150 7783 A0D1 7507 1266  C5B3 BBF1 9C42 1E32 0E65

I don't like the idea of secret agencies to analyse and archive
personal communication. GnuPG is available as open source, free as as in
freedom, as a countermeasure. I use http://www.enigmail.net/ for Mozilla
Thunderbird. If you can, please use a frontend of your choice to send me
encrypted e-mail. See http://www.gnupg.org/ for an overview.



signature.asc
Description: OpenPGP digital signature
--
Dive into the World of Parallel Programming! The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users


Re: [Bacula-users] btape fill failure with LTO-3 on FreeBSD 9.3

2015-01-14 Thread Georg Altmann
Am 14.01.2015 um 00:52 schrieb Dan Langille:
> 
>> On Jan 4, 2015, at 5:05 PM, Georg Altmann  wrote:
>> The prompt
>>
>> Mount Volume "TestVolume1" on device "LTO3-0" (/dev/nsa0) and press
>> return when ready:
>>
>> just re-appears every time I hit enter. I have attached the full btape
>> output.
>>
>> What about the error message "Error: mount.c:834 Hey! WroteVol
>> non-zero !" ?
>>
>> This is the bacula-sd configuration of the device:
>>
>> Device {
>>  Name = LTO3-0
>>  Media Type = LTO3
>>  Archive Device = /dev/nsa0
>>  AutoChanger = no;
>>  Spool Directory = /bspool/bacula
>>  AutomaticMount = yes;   # when device opened, read it
>>  AlwaysOpen = yes;
>>  RemovableMedia = yes;
>>  RandomAccess = no;
>>  Hardware End of Medium = no;
>> }
> 
> Your configuration is interesting.  Is this a standalone tape drive?

Yes, this is a standalone drive.

>>
>> This comes as a bit of a surprise to me, since I successfully operated
>> an HP LTO-1 drive with the same configuration on FreeBSD successfully.
>> However, this was a previous version of FreeBSD and bacula.
>> Operating the drive with tar and mt (fsf) works just fine.
>>
>> Might this just be a regression with btape?
> 
> When you ran your successful backup and restore spanning two LTO3 tapes, did 
> you do a diff on the original version?

Yes, I md5'ed the files before the backup and after the restore and
diffed the checksums. The checksums were identical.

I am quite convinced that that configuration I have is fine and that
this is a btape problem.
The drive is now in productive operation (remote!) and I don't want to
mess with the config. I might do some tests once I am on location again.
To my understanding, it would make sense to look into how btape differs
from bacula-sd in handling the drive.

As said before, this configuration has worked just fine for LTO1 drives
on FreeBSD with and without an autoloader. I am guessing that the
SCSI/IOCTL interface hasn't changed between the LTO generations.

Thank you for your input, Dan!

Regards,
Georg

-- 
PGP-Key: 0x1E320E65
D150 7783 A0D1 7507 1266  C5B3 BBF1 9C42 1E32 0E65

I don't like the idea of secret agencies to analyse and archive
personal communication. GnuPG is available as open source, free as as in
freedom, as a countermeasure. I use http://www.enigmail.net/ for Mozilla
Thunderbird. If you can, please use a frontend of your choice to send me
encrypted e-mail. See http://www.gnupg.org/ for an overview.



signature.asc
Description: OpenPGP digital signature
--
New Year. New Location. New Benefits. New Data Center in Ashburn, VA.
GigeNET is offering a free month of service with a new server in Ashburn.
Choose from 2 high performing configs, both with 100TB of bandwidth.
Higher redundancy.Lower latency.Increased capacity.Completely compliant.
http://p.sf.net/sfu/gigenet___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users


Re: [Bacula-users] Offsite Storage

2015-01-26 Thread Georg Altmann


Am 23.01.2015 um 09:27 schrieb Uwe Schuerkamp:
> On Thu, Jan 22, 2015 at 06:30:01PM +, Damien Hull wrote:
>> I've asked this question before, but I don't think I got enough information. 
>> Here's my situation
>>
>> 1.  I have three offices in different physical locations.
>>
>> 2.  Each location has between 200 GB to well over 1 TB of data.
>>
>> 3.  I would like to backup data between offices for offsite storage.
>>
>> 4.  I can't wait weeks for data to be copied between offices
>>
>> I know some commercial software allows you to "seed" the backup.
>>
>> 1.  Copy data to an external drive
>>
>> 2.  Mail the drive to another office
>>
>> 3.  Copy the data over - Your backup is now "seeded"
>>
>> 4.  Backups are now faster because you don't have to copy everything.
>>
>> Can bacula handle this?
>>
>> Thanks!
>>
> 
> While I'm not entirely sure bacula could handle *all* of this, I'm
> pretty certain you can come up with a clever solution based on rsync
> and bacula.
> 
> It depends on how much data changes on all of the servers involved,
> but I'd imagine the following setup would be possible (given that disk
> space is cheap):
> 
> - seed the backups using a mailed drive / tape on local storage,
>   ideally connected to your local bacula server
> 
> - rsync changes nightly over to the offsite locations (runbeforjob?)
> 
> - backup your local file system copies on a local library at the
>   offsite locations, using the local bacula instance
> 

Why would you want to use rsync for that?

bacula can do base jobs (see the bacula manual). You can use them as a
"seed" backup. I.e. make a base backup to data media of your choice and
mail them to the remote site.
Then you can do remote full/differential/incremental backups using a
remote bacula sd over a secure connection (ssh, vpn). If full backups
are too big to do them over the net, you might want to mail these as well.
For managing local and remote media, I think two pools associated with
the local and remote storage devices would do the job. After mailing the
remote media you would then move them from the local to the remote pool.

Of course you can add copy jobs to keep local and remote copies of the
backups.

Regards,
Georg

-- 
PGP-Key: 0x1E320E65
D150 7783 A0D1 7507 1266  C5B3 BBF1 9C42 1E32 0E65

I don't like the idea of secret agencies to analyse and archive
personal communication. GnuPG is available as open source, free as as in
freedom, as a countermeasure. I use http://www.enigmail.net/ for Mozilla
Thunderbird. If you can, please use a frontend of your choice to send me
encrypted e-mail. See http://www.gnupg.org/ for an overview.



signature.asc
Description: OpenPGP digital signature
--
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users