limit incremental backup to list of changed files

Asked by Thomas Tanner

Hi,
I'd like to speed up incremental backups by using a list of created/modified/deleted files generated by fswatch
http://emcrisostomo.github.io/fswatch/

Is there a simple way to specify this file list?
AFAIK --include-filelist would delete all unchanged files from the backup in this case.

thank you

Question information

Language:
English Edit question
Status:
Answered
For:
Duplicity Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Aaron Whitehouse (aaron-whitehouse) said :
#1

Hi Thomas,

I'm pretty sure that this is not yet possible, but it is a feature that I have been considering for some time. There was some talk about it back in 2010:
https://mail.gnome.org/archives/deja-dup-list/2010-February/msg00004.html

I was similarly looking to provide a list of changed files from inotify or similar. It would make implementing real-time monitoring in programs that use duplicity much easier, too, for instance Bug #781428

It probably should be added as a feature request.

Revision history for this message
edso (ed.so) said :
#2

On 28.01.2015 20:51, Thomas Tanner wrote:
> New question #261358 on Duplicity:
> https://answers.launchpad.net/duplicity/+question/261358
>
> Hi,
> I'd like to speed up incremental backups by using a list of created/modified/deleted files generated by fswatch
> http://emcrisostomo.github.io/fswatch/
>
> Is there a simple way to specify this file list?

no, currently not ;(

> AFAIK --include-filelist would delete all unchanged files from the backup in this case.
>

yes, it would.

usually the slow part with duplicity isn't the backup creation, but it's transfer to the backend. did you at some point compare backup times from the same source to either local file and a remote location, to get a feel?

problem is that duplicity is not really multithreading, so it waits for transfers to succeed and resumes scanning thereafter until it has enough diff for the next volume. the asynchronous option does tackle this, but only for one volume scanning, while the previous is still transmitting.

..ede/duply.net

Revision history for this message
Thomas Tanner (tomtanner) said :
#3

Identifying the modified files might not be the most time consuming part,
but even scanning the whole directory tree of a large filesystem (e.g., find) may take a long time.
Using fswatch would give you the result instantly.

Can you help with this problem?

Provide an answer of your own, or ask Thomas Tanner for more information if necessary.

To post a message you must log in.