Bacula-users

Re: [Bacula-users] Some questions about bacula

2011-02-18 18:01:40
Subject: Re: [Bacula-users] Some questions about bacula
From: Frank Sweetser <fs AT WPI DOT EDU>
To: Mike Hobbs <mhobbs AT mtl.mit DOT edu>
Date: Fri, 18 Feb 2011 17:57:35 -0500
On 2/18/2011 2:15 PM, Mike Hobbs wrote:
>    Hello,
>
> I'm a very new Bacula user and am looking for a little help.  My
> experience for the past 10+ years has been with Amanda and Retrospect
> and I've finally had enough of both (especially Retrospect) and I'm
> looking to convert to Bacula.
>
> I've download the Bacula source and built in on Red Hat 5.5 x64.  I'm
> also trying to move to a disk-based backup solution and I'm using a
> Promise Technologies VessJBOD 16-bay populated (or will be) with 16x 2tb
> hard drives.
>
> Without editing the config files to much I have done successful backups
> and restores of both Linux and Windows machines.  So far so good.  I
> have to say disk based backups and restores are FAST!

Sounds like you're off to a good start so far =)


> Some questions I have now that I'm ready to start customizing Bacula
> more for my environment.
>
> 1)  Bacula, like Amanda, needs to be told what machine and file systems
> to be backed up.  I understand that you add clients (jobs)  to the
> bacula-dir.conf file.  This would be fine if I had only a few clients.
> What do I do when I need to add 50 or 100 clients?  Is there a way to
> auto-generate the list of machines and file systems to be backed up
> every night?  On Amanda one of the programmers on staff wrote me a
> script that basically pings all the clients and looks at them to see
> what file systems are mounted, then it generates the Amanda "disklist"
> file based on what it finds.  This is good because if a user decided to
> change the name of a file system and doesn't tell us, the script will
> see it and add the new file system to the list and we don't have to
> worry about editing the disklist file manually..  Is there some way to
> do this with Bacula?  I hope I made sense here.

Yes, absolutely.  You can set up the director config to include the output of 
an 
arbitrary script at any point in the config file:

http://bacula.org/5.0.x-manuals/en/main/main/Customizing_Configuration_F.html#SECTION001723000000000000000

> 2) Here is a quote from an email of what one of my co-workers would like
> to do with bacula, I believe this is doable using "Pools" but if someone
> could point me in the right direction.
>
> "If we do a setup similar to Retrospect, then we want two sets of equal
> size.  It should do each set for 6 weeks and it should continue doing
> incrementals for the first set while it's transferring over to the next
> set and doing level0s - assuming this is supported by bacula.  At any
> time, we always have at least 6 weeks of backups but most times, we'll
> have something between 6-12 weeks."

I'm not at all familiar with Retrospect, but I'd be careful you don't try to 
recreate Retrospect in Bacula, rather than making sure you're meeting your 
goals.  In this case, it looks like you're trying to ensure that any data 
stored 
on tape is kept for a minimum of 6 weeks, correct?  If so, you can do that with 
a single pool by setting up your retention periods properly.  I'd advise you to 
read the chapters on retention periods very carefully, as they're easy to get 
tripped up on.

> 3) As I mentioned I have JBod unit with 16x 2tb drives.  I'm thinking 8
> drives for the first "set" and the other 8 drives for the second "set".
> If the about example of what my coworker wants to do is allowed with
> bacula.  How many volumes would I want to create per drive?  And what
> would be the max size?  I read a thread on the bacula-mailing list,
> someone said you want 10-100 volumes per disk.  I'm not sure what would
> be the best way to determine how many volumes I will need.

Again, you'll probably be fine with a single pool and the right retention 
periods.  One factor to be aware of, though, is that when Bacula recycles a 
volume, it recycles everything on that volume all at once - all jobs and all 
files on it are wiped from the database, and subject to being overwitten.  As a 
result, you don't want gigantic volumes, which would make recycling too coarse. 
  A typical recommendation is around 10-50 gigs, depending on your backup needs.

> I'm sorry if these questions have been asked already or if I left out
> any information.. I did search the list for a little while without
> finding any results.
>
> I guess I'm looking for some kind of disk-based backup strategy for
> about 100 Linux machines with the possible addition of another 100
> Windows machines in the future.

It sounds like you're definitely on the right track.

-- 
Frank Sweetser fs at wpi.edu  |  For every problem, there is a solution that
WPI Senior Network Engineer   |  is simple, elegant, and wrong. - HL Mencken
     GPG fingerprint = 6174 1257 129E 0D21 D8D4  E8A3 8E39 29E3 E2E8 8CEC

------------------------------------------------------------------------------
The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
http://p.sf.net/sfu/intel-dev2devfeb
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

<Prev in Thread] Current Thread [Next in Thread>