[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Duplicity-talk] Basic support for glacier in bzr
From: |
edgar . soldin |
Subject: |
Re: [Duplicity-talk] Basic support for glacier in bzr |
Date: |
Mon, 21 Jan 2013 11:23:48 +0100 |
User-agent: |
Mozilla/5.0 (Windows NT 5.1; rv:16.0) Gecko/20121026 Thunderbird/16.0.2 |
On 21.01.2013 11:13, address@hidden wrote:
> On 21.01.2013 02:06, Eric Friedrich wrote:
>> Hello all
>>
>> I am a big fan and user of duplicity but have not contributed before. I'm
>> very interested in using Amazon Glacier as my storage service but without
>> the hassle of passing my data through S3.
>>
>> I've committed very basic support for glacier to bazaar. Unfortunately
>> glaciers retrieval times are slow and back end gets are serialized so the
>> whole restore process can take a very long time.
>>
>> I'd appreciate any suggestions to improve the diff, especially how to
>> parallelize the gets.
>>
>
> hey Eric,
>
> you have a look at what duplicity does on a restore. but essentially it
> downloads one volume, works with it, downloads the next, works with it...
> currently there's no way of preloading anything, by the looks of it a design
> choice.
>
btw. i have been thinking about implementing a duplicity 'copy' command which
would simply reuse the broad backend support in duplicity to copy files from
one backend to another.
if you'd hack such a thing you could copy your repository to local and restore
from there. if you want to go a step further you could implement the --time
switch so that you could limit the chains that will be downloaded.
of course this will not cherry pick volumes with data for one file restores,
but it'd be a start.
..ede/duply.net