[PATCH] CTDB protocol and client fixes

[Volker Lendecke]

On Tue, Nov 24, 2015 at 05:33:23PM +1100, Amitay Isaacs wrote:
> Since no one has replied to this email, let me summarize.
> 
> 1. For some reason I cannot fathom, async processing is treated *different*
> from sync processing.  A global event context for all async processing is
> ok, but for sync processing need lots of temporary contexts.

When you use tevent_loop_once deep inside the call stack where some
other tevent_loop_once is also active, anything can happen. As long as
callers work without tevent_loop_allow_nesting then it might be workable.

> 2. I can drop the tevent_context from sync API, but it will still use the
> cached event context rather than creating a new one for every sync API
> call.  This is because of the current design of the client API.  It was
> designed to not use tevent_add_fd() for reading each individual packet from
> an fd.

"cached event context" is something that I am worried about using in
smbd and winbind.  We have gone through a lot of work to remove this from
the core smb libraries. You might prove me wrong by finding spots where
we are not there yet, but in the SMB libraries I would see that as
a bug that needs fixing.

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de