[Gluster-devel] Re: New wiki page for 2 server afr, client side afr

[Krishna Srinivas]

On Fri, May 2, 2008 at 11:33 AM, Brandon Lamb <brandonlamb at gmail.com> wrote:
> On Thu, May 1, 2008 at 10:49 PM, Krishna Srinivas <krishna at zresearch.com> wrote:
>  > Brandon,
>  >
>  > $ echo "hello" > file.txt
>  > -bash: file.txt: Input/output error
>  >
>  > Can you check in the logs if there is something related to this?
>  > It should have worked fine there.
>  >
>  > Also for client2, the order should be "brick2 brick1" for AFR
>  > (it should always  be same for all clients for server2, brick2 connects
>  > to brick1 of server1)
>  >
>  > Krishna
>  >
>  >
>  > On Fri, May 2, 2008 at 3:21 AM, Brandon Lamb <brandonlamb at gmail.com> wrote:
>  > > On Thu, May 1, 2008 at 2:47 PM, Brandon Lamb <brandonlamb at gmail.com> wrote:
>  > >  > http://www.gluster.org/docs/index.php/Setting_up_AFR_on_two_servers_with_client_side_replication
>  > >  >
>  > >  > Look over and make sure it is kosher?
>  > >  >
>  > >  > I added a section at the bottom for "gotchas", can you take a quick
>  > >  > look to make sure they are accurate statements.
>  > >  >
>  > >  > =P
>  > >
>  > >  I also just added a page for server side replication
>  > >
>  > >  http://www.gluster.org/docs/index.php/Setting_up_AFR_on_two_servers_with_server_side_replication
>  > >
>  > >  i'll start filling that in now too
>
>  Oh hm, how does the order of the subvolumes listed affect it? I wasnt
>  aware that it made any difference, I thought you could list subvolumes
>  brick1, brick2, brick4, brick3 in any order.
>

It is used internally by AFR for locking, always the first server is used for
locking. We need to lock when changing/comparing versions of the files
and directories during selfheal.

Krishna