[OpenAFS] "vos release" improvement proposals

Horst Birthelmer horst@riback.net
Wed, 16 Nov 2005 18:43:51 +0100

On Nov 16, 2005, at 5:44 PM, Peter Somogyi wrote:

> Hi,
> I'm working on improvements on syncing sites ("vos release").
> Currently I'm implementing rsync into volserver. I'm in a hope it  
> will be a good speedup.
> And I hope this feature will be accepted into CVS (after when it's  
> ready and works). Please tell me if not.
> I have further plans for improvements:
> - use a different, own protocol specially for "vos release"  
> replacing the current one, branching from "vos dump"
> (When syncing, currently the same data is sent across the network  
> which results in a "vos dump".
> I think this dump format is too generic & redundant, and we could  
> gain extra speedup and less data to send for small files when the  
> protocol would be fixed, omitting "tag"-ing.)
The dump format is just right...
I think your real improvement and valuable contribution should be a  
new protocol just for the replication release.
Which means the only paradigm you should be breaking is the format  
used by the volume release, _not_ dump and restore.

Well there are some 'historical' assumptions made in the dump format,  
but I don't think that has to be changed right now, even though this  
kind of ideas are 'floating' around in the community for quite some  
time now.

Wouldn't that create the possibility of determining the difference  
between two files (not volumes)?
If yes, how do you deal with different versions on different sites.  
Which lets you have n different versions of a file (n being the  
number of RO volumes)?

> - profile & optimize "vos release": if anybody have experience when  
> this "vos release" gets extraordinarily slow and knows the  
> environmental reason please write.

AFAIK it's transmitting all changes to the volume but on file basis,  
which generates a lot of traffic for small changes to large files.


P.S.: just my two coins ;-) trying to make sure you don't change the  
volume dump and restore parts unnecessarily