CTDB and the road ahead.

[Amitay Isaacs]

Hi Chris,

On Tue, Jan 15, 2013 at 6:09 AM, Christopher R. Hertel <crh at ubiqx.mn.org>wrote:

> Amitay, et. al.,
>
> I need to get a sense of where CTDB is going.  My initial understanding of
> the 2.0 release was that it was meant to pull together a number of
> disparate patches and become the solid foundation for further development.
>  When I speak to different people, however, I get different feedback.  I
> also understand that there may be some personal repositories scattered
> about.
>
> So I'm confused.  Help me (and Samba vendors and users) understand at a
> basic level what the plans are and which code base we should be using.
>

CTDB 2.0 was mainly to abandon the old and confusing numbering scheme. So
CTDB branch 1.13 (CTDB version 1.44) is now completely defunct and should
not be used. I would prefer to do regular releases of CTDB 2.x (time
permitting). Currently 2.x releases are done from master which makes it
easier to manage.

As for the direction of CTDB, there are plenty of things to do. My current
focus is on identifying race conditions and fixing them. For some of the
workloads CTDB does not perform very well, so will continue to work on
those. Currently there is no well defined libctdb for client applications
to use, so work is needed in that area as well. Also if we can figure out a
way of running samba autobuild tests on a CTDB cluster, that would improve
the integration (ctdb+samba) testing considerably. These are just some of
the areas. May be others can identify more.

So as you can see there is lots to do. :-) Any contributions are welcome!


> Thanks.
>
> Chris -)-----


Amitay.