Re: [Gluster-devel] New IDEA: The Checksumming xlator ( AFR Translator have problem )

[Gareth Bult]

Yes, that is correct .. but there are many instances when you want to restart a 
server .. let alone in the case of a server failure or upgrade ...


----- Original Message -----
step 3.: "Sascha Ottolski" <address@hidden>
To: address@hidden
Sent: 17 January 2008 11:12:07 o'clock (GMT) Europe/London
Subject: Re: [Gluster-devel] New IDEA: The Checksumming xlator ( AFR Translator 
have problem )

Am Donnerstag 17 Januar 2008 09:47:06 schrieb Angel:
> But the real problem as you pointed out is that AFR only know files
> dont match, so have to copy every byte from you 100GB image from B to
> A (automatically on self-heal or on file access )

i never tried it, but isn't this a situation where rsync's "--inplace" 
parameter is useful? seems to be able to kinda binary diff on a 
blocklevel. may such an approach could be used for afr self-heal as 
well?

btw, just curious, the problems with xen do only occur after a node was 
down, right? if i understand it correctly, as long as all afr nodes are 
up the client would write to them in parellel, so that the nodes stay 
in sync automtically at any time?


Cheers, Sascha



_______________________________________________
Gluster-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/gluster-devel