On Mon, Apr 13, 2009 at 6:02 PM, James Chamberlain <jamesc AT exa DOT com> 
wrote:
>>> Why would you ever want such a pool?  The only reason I can think of is
>>> if
>>> you have more pools than backup devices;
>>
>> Exactly what you said. I have 20 pools and 2 backup devices with my 2
>> drive 24 slot autochanger.
>
> Why so many pools?  Are you doing one per client?
>
Not exactly. More closely to 1 pool per type of data. Additional pools
were added for independent duplicates of the data.

>>> but that's the opposite of the
>>> problem I'm trying to solve.  I have more backup devices than pools.  In
>>> some sense, I want to have multiple devices within the same pool.
>>>  Ideally,
>>> I'd like to have one of those devices in multiple pools.  I want the
>>> volumes
>>> and not the devices bound to the pool.  Each pool could then tag any
>>> volumes
>>> it uses with the correct pool label and return them to scratch when they
>>> expire.  That's that I was hoping for when I read the documentation for
>>> the
>>> scratch pool, though that interpretation is apparently incorrect.
>>>
>>> The basic problem for me is that I've hit the 8 TB file system size limit
>>> with ext3, and I don't have ext4 available to me yet.
>>>
>> I would use XFS over ext3. ext3 is horribly inefficient with large files.
>
> That's not something I can change at this point.
>
I understand.

>
>>> With tape libraries,
>>> you can keep adding more tapes to increase the size of your pool.  With
>>> disk-based backups, once you've hit that 8 TB limit with ext3, you can't.
>>>  So if that's the problem I'm trying to solve, what are my options with
>>> Bacula?
>>>
>>
>> Can you split your jobs up in some logical way so you can divide the
>> storage in more than 1 part?
>
> I think I have.  I have two 8 TB file systems and one 1.1 TB file system.
>  My individual volumes on each of those file systems are only 50 GB.  I just
> have 160 of them per 8 TB file system and 20 on the 1.1 TB.  I have another
> 1.1 TB left over that I want to use if any of the others fill up before old
> jobs expire.  As for splitting up the jobs, I've broken them into desktop,
> server, and infrastructure pools.  Is that what you were getting at?
>

Yes.


The other option will take work/research from you. The virtual disk
autochanger concept. There has been discussion about this on the list.
The theory is to create a diskchanger script similar to the mtxchanger
script that make the individual arrays look like just 1 storage device
or 1 auto changer (with 1 or more storage devices) then with dozens or
even hundreds of volumes and the ability to switch mount points to the
different physical arrays to access volumes in the virtual slots. I
hear mention of a virtual disk changer in bacula 3.0.0, however I have
not investigated that.

John

------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users