Re: [Duplicity-talk] duplicity 0.7 slowness

[Philip Jocks]

Hej,

> does it keep being fast after recreating the archive dir?

you mean on an empty archive dir? Yes, still fast, the working directory (hash 
of the target URL) inside the archive dir stays empty.

> 
> ..ede/duply.net
> 
> PS: Aaron reworked file selection lately. another user came up w/ a patch 
> that accelerates duplicity by magnitudes, maybe you want you try it? 
> https://code.launchpad.net/~mwilck/duplicity/0.7-series/+merge/301332

yes, I just tried it (replaced globmatch.py and selection.py), but nothing is 
gained, collection-status still takes >50sec with —name vs. <1sec without —name.

Does —name have any other purpose other than having human readable working 
directories in the archive-dir vs the hash of the target URL? For me, leaving 
out —name would be fine, duplicity even mentions the directory while running, 
maybe adding a switch to duply might even suffice for the time being.

Am I the only one seeing this or is anyone else able to reproduce it? Currently 
testing it with a backup set of 300+ volumes of 200MB each, it might not be so 
sever with smaller sets, though it seems it isn’t even a matter of file size 
since it seems to only list difftar files etc, no real gpg magic etc.

Cheers,

Philip