[ovirt-users] two node replicated gluster

[Vijay Bellur]

On 09/25/2014 07:44 PM, ml ml wrote:
> Hello List,
>
> i have a Two Node replicated Gluster.
>
> I am running about 15 vm on each host so that in case one node fails
> the other one can take over.
>
> My question is how the GlusterFS self-healing-daemon works.
>
> I disconnected the two nodes on purpose and reconnected them again. So
> that the self-heal-daemon jumps it.
>
> It looks like it would transfer all the file. Does the
> self-heal-deamon of glusterfs not just copy the missing increments?
>

self-heal daemon just syncs regions of files that need to be updated and 
the process of identification of regions to be updated involves 
comparison of a rolling checksum of various regions in the files. You 
can find more details about Gluster's replication and the self-healing 
algorithm in the "Recovering from failures (Self-heal)" section of this 
document [1].

-Vijay

[1] https://github.com/gluster/glusterfs/blob/master/doc/features/afr-v1.md