Amanda-Users

Re: setting up a dump cycle

2004-01-16 12:56:51
Subject: Re: setting up a dump cycle
From: Jon LaBadie <jon AT jgcomp DOT com>
To: amanda-users AT amanda DOT org
Date: Fri, 16 Jan 2004 12:53:26 -0500
On Fri, Jan 16, 2004 at 06:10:09PM +0100, Eugen Leitl wrote:
> Jon LaBadie wrote:
> 
> >
> >
> >Have you read the exclude how-to in the <amanda_source>/docs directory?
> 
> No, I haven't -- thanks for the pointer. I was just aware that there's a 
> thing
> as an exclude list. I'm using GNU tar, so there shouldn't be a problem.
> 
> I'm wondering about
> 
> bash-2.00$ df -k
> Filesystem           1k-blocks      Used Available Use% Mounted on
> /dev/dsk/c0t0d0s0       385463    257438     89479  75% /
> /dev/dsk/c0t0d0s6       962571    507261    397556  57% /usr
> /dev/dsk/c0t0d0s3       385463    246616    100301  72% /var
> /dev/dsk/c0t8d0s7     17413250  14751224   2487894  86% /Disk2
> /dev/dsk/c0t0d0s7      8335201   6781694   1470155  83% /export/home
> /dev/dsk/c0t0d0s5      6050182   4547150   1442531  76% /opt
> /dev/dsk/c0t0d0s1       770943    310513    406464  44% /usr/openwin
> swap                    271088      2792    268296   2% /tmp
> 
> though, this shouldn't fail with a DDS3 tape (this is all pretty
> compressible data). Maybe I haven't resent the cycle probably.

Well, do you have HW compression on?  And software compression too?
In that case the SW compressed data will expand during HW compression.
Probably not beyond the original data size if that was reasonably
compressible in the first place.  But ...

If that df cmd represents your DLE's too, then you might want to
consider splitting with the exclude/include features at least
the /Disk2 FS.  The reason being balance from day to day.  If that
were split into say 3 or 4 chunks in your disk list, then amanda
would have 5 or 6 big DLE's to spread the level 0's amongst the
dumpcycle.  Rather than little today, little tomorrow, BANG /Disk2
the next day.

BTW on Solaris (looks like Solaris from the above) and HP DDS3 tapes,
only the "ell" devices (/dev/rmt/*l*) are sure to be non-compressing.

> Is
> amlabel -f daily DailySet14
> amlabel -f daily DailySet15
> etc.
> sufficient to simulate presence of a new tape, or need I
> actually insert a blank tape, or to wipe the existing one
> (how?).

Not sure what your objective is.  In your original posting you said
something about "restarting a dumpcycle".  For normal amanda use,
every day is the start of a dumpcycle.

In case you or I am misunderstanding things, you can't amlabel without
a tape in the drive.  A blank, new, unlabeled tape can not be used by
amanda.

If you mean you want to start production backups using the same config
as your test bed ...  it has been a while but at one time I was doing
that regularly for testing.  I had a script that set the config back
to a known initial state.  I think it included

  editing the tapelist file to put it in totally unused state,
  IIRC this was extract the middle column, sort descending, add
  a column of 0's at the beginning of each line and a column of
  "reuse" at the ends.

  rm -rf of the curinfo/*, index/*, and logdir/* (note I specify
  logdir as a subdirectory of my config much like index and curinfo.
  I don't think that is the default.

  maybe also I did something with the gnutar list directory, but
  I don't think so.

But be aware that this will cause the entire set of DLE's to get
level 0's the same day.  Massive backup time and tape(s).  Maybe
better to comment out most of disklist, uncommenting a few each
subsequent day.

Other less drastic ways would require reading the man pages for
things like amadmin force level 0's, or amrmtape, or ???

jl
-- 
Jon H. LaBadie                  jon AT jgcomp DOT com
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)

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