[Gluster-devel] Server-side AFR

[Gordan Bobic]

Krishna Srinivas wrote:
> You can check the foll wiki:
> http://www.gluster.org/docs/index.php/GlusterFS_1.3_High_Availability_Storage_with_GlusterFS
> 
> Though it is not entirely correct (unify does not serve any purpose in that
> example) you will get an idea on how to use AFR on the server side.
> Check how "mailspool-ds-afr" is defined.
> 
> You can mail back if it still confusing.

I got it working, thanks. :)

Now I seem to be getting the same problem that Guido Smit described in 
another thread. When one node gets shut down, trying to access the mount 
point on the other node blocks, seemingly indefinitely.
killall glusterds; killall glusterfsd;
and restarting seems to clear the problem, even though the other node is 
still gone.

Another question - if I want to "move" a volume to GlusterFS, do I have 
to copy files to the GlusterFS mountpoint? Or can I just export the 
already populated file system as the source directory of the posix 
storage block?

e.g. if I want to migrade /home to be GlusterFS exported, can I do just 
do something like:
cd /
mv home gluster
mkdir home
glusterfs -f /etc/glusterfs-client.vol /home

provided that glusterfs-client is using /gluster as the data source? Is 
there any reason why this wouldn't work? Or do I really have to copy all 
the content to the mounted GlusterFS the long way?

Thanks.

Gordan