|
Re: how to split partitions
2006-04-03 12:18:25
Paul Bijnens wrote:
On 2006-04-03 07:38, Josef Wolf wrote:
On Sun, Apr 02, 2006 at 09:34:59PM +0200, Paul Bijnens wrote:
That would mean that the estimate phase should be smart enough (may
be add an option "autosplit" for those DLE's where you want that) to
pick a few large subdirectories out and then split the DLE dynamically.
This is not what I have meant. No autosplitting. Just automatically
create appropriate exclude entries. When one sees a DLE starts becoming
too big, just pich one of its subdirs and create a new DLE for this
directory, that's all. No fiddling with exclude list entries any more.
I understood that after your last mail.
The problem with that approach is that the Amanda server does not
know if a subdir is on another mountpoint or not.
It only needs excludes if it is on the same mountpoint.
After my coffee, I'll understand the implications of adding it always,
even if the directory is on another mountpoint:
mounts:
/dev/hda1 on /
/dev/hda3 on /var
with disklist:
pluto / comp-tar
pluto /var comp-tar
pluto /var/spool comp-tar
getting excludes similar to:
pluto / {
comp-tar
exclude append "./var"
}
pluto /var {
comp-tar
exclude append "./spool"
}
pluto /var/spool comp-tar
Nah, this is way too complicated!
While working on the setup I did, it wasn't particularly difficult to
get the entries right. What caused me the most trouble after getting to
the idea how to do it, was only not knowing exactly how things (like
exclude/include, specifying disk names and paths relative to
directories) would work. Once that has become clear, the disk list is
rather lengthy, but not complicated. I actually think it is quite
elegant and beautiful the way it is.
As for helping users in creating such a disklist, I can imagine a
program that would be supplied with few and simple parameters like the
root of some file system and a level specifying how deep it should look
into the directory structure for splitting, starting from the root.
The program would just put out a disk list (part of a disk list) to
stdout, containing appropriate entries with include and exclude
statements as needed.
In a next step, the program might as well consider the size of the
resulting dumps and eventually suggest using a deeper level of splitting.
The program could be used to play around a little and examine the file
systems that are to be backed up. Its output could be concatenated to
create the actual disklist. That disklist would remain static. Since
users still build the disklist themselfes, they will still know what's
going on and keep things under their control. But they would have some
help with it.
GH
|
|
|
