[Gluster-devel] Weird AFR behaviour with combination of local storage bricks and remote volumes

[Anand Avati]

Sam,
 could you please try
http://gnu.zresearch.com/~avati/glusterfs-1.3.8tla607.tar.gz and see if you
still have the same behaviour?

avati

2007/12/12, Sam Douglas <sam.douglas32 at gmail.com>:
>
> Hello
>
> With both 1.3.7 and TLA:
>
> I have a test AFR setup along the lines of:
>
> ###
>
> volume brickA
>     type storage/posix
>     option directory /stuff/brickA
> end-volume
>
> volume clientA
>     type protocol/client
>     option transport-type tcp/client
>     option remote-host 10.0.0.1  # On some other system
>     option remote-subvolume brickA
> end-volume
>
> volume afrA
>     type cluster/afr
>     option replicate *:2
>     subvolumes brickA clientA
> end-volume
>
> volume server
>     # exports brickA and afrA
> end-volume
>
> ###
>
> When a client mounts the AFR volume, files are not created on the
> storage/posix volume, only on the client volume. Symlink, mknod etc.
> still work.
>
> Adding a debug/trace over the storage/posix shows that lookup and symlink
> fops
> are being called on the brick, but no readv/writev and open calls.
>
> If another protocol/client translator is added that connects to the
> exported storage
> volume via the server, then everything works as expected.
>
> Any clues?
>
> -- Sam Douglas
>
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at nongnu.org
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>



-- 
If I traveled to the end of the rainbow
As Dame Fortune did intend,
Murphy would be there to tell me
The pot's at the other end.