Amanda-Users

Re: tape_splitsize Problem: Resulting Splits Are Too Small

2007-12-07 10:30:15
Subject: Re: tape_splitsize Problem: Resulting Splits Are Too Small
From: Charles Curley <charlescurley AT charlescurley DOT com>
To: Marc Muehlfeld <Marc.Muehlfeld AT medizinische-genetik DOT de>
Date: Fri, 7 Dec 2007 08:21:30 -0700
On Fri, Nov 16, 2007 at 08:37:10PM +0100, Marc Muehlfeld wrote:
> Charles Curley schrieb:
> > The dump is to cirtual tapes on a hard drive. The resulting splits are
> > all 10 MB in size. E.g.:
> >
> > How do I change that to a more reasonable size, like 3 Gb?
> 
> The parameters fallback_splitsize default is 10M, if you don't set it.
> Maybe the conditions descripted inside the manpage of amanda.conf appears
> in your case and let amanda fall back to use this parameter.

That seems to have helped a lot. Amanda finally backed up again the
same DLE that brought up the issue. Unfortunately, I set the value to
1 Gb, and got two splits, 1.7G and 3.0G.


> 
> 
> fallback_splitsize
> Default: 10M When dumping a split dump in PORT-WRITE mode, if no
> split_diskbuffer is specified (or if we somehow fail to use our
> split_diskbuffer), we must buffer split chunks in memory. This specifies
> the maximum size split chunks can be in this scenario, and thus the
> maximum amount of memory consumed for in-memory splitting. The size of
> this buffer can be changed from its (very conservative) default to a value
> reflecting the amount of memory that each taper process on the dump server
> may reasonably consume.

Sorry, but even after reading your remarks above and experimenting
with it, this writeup is still obtuse. For one thing, I had no idea I
was splitting in memory. Isn't that normally done on the holding disk?


-- 

Charles Curley                  /"\    ASCII Ribbon Campaign
Looking for fine software       \ /    Respect for open standards
and/or writing?                  X     No HTML/RTF in email
http://www.charlescurley.com    / \    No M$ Word docs in email

Key fingerprint = CE5C 6645 A45A 64E4 94C0  809C FFF6 4C48 4ECD DFDB

Attachment: signature.asc
Description: Digital signature

<Prev in Thread] Current Thread [Next in Thread>
  • Re: tape_splitsize Problem: Resulting Splits Are Too Small, Charles Curley <=