[EMAIL PROTECTED] schrieb:
> Tobias Sarnowski:
>   
>> Index | Mode
>>     0 | read-only
>>     1 | read-only
>>     2 | writeable
>>     3 | read-only
>>
>> How I understand the documentation (see P.S.), I would expect the
>> following behaviour:
>>
>> 1. I write something to a file. AUFS will go down to index 2 and write
>> the content to the file
>>     
>
> At this point, does the file already exist on index 0 or 1 readonly branch?
> If it is true, you won't write to the file.
>   
And here is my problem. The file *could* exist on the higher readable
branch, but should be ignored then. To avoid that, I want to reorder the
AUFS index but do not found a good practice.

Following up I have the questen, if a directory can be mounted twice or
more as different branches? I could add the writeable directory on top
again without the need to unmount the first branch of it.
> If it is not, does the parent dir already exist on index 0 or 1 readonly
> branch? In this case, you won't write either.
>
> If none exists on index 0 and 1, and the parent dir exists on index 2,
> you can write and read to/from the file on index 2.
> (I don't assume you are bypassing aufs. In other words, you always
> access the file through aufs instead of directly to the branches.)
>
>
>   
>> 2. Afterwards I want to read that file (and here comes my problem) -
>> AUFS gives me the version of index 0 because it is the highest one.
>>     
>
> When the same named file exists on index 0, it is an expected behaviour.
>
>
>   
>> If it gives me the newer version of index 2, what is, if I mount a new
>> branch on top of all (index 0) and reread the file. Do I get the index 2
>> version (then 3) or the index 0 one?
>>     
>
> When the file exists on index 0, you will get the one on index 0.
>
>   

-- 
Tobias Sarnowski / software development
__ CosmoCode GmbH - http://www.cosmocode.de
__ Prenzlauer Allee 36, 10405 Berlin, Germany
__ Tel +49-30-28096200, Fax +49-30-28097093
__
__ Geschaeftsfuehrung (CEO): D. Huettemann, J. Riebesell
__ Amtsgericht Berlin-Charlottenburg; HRB 74063 


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Reply via email to