Work on SMB3 persistent handles

[Christopher R. Hertel]

On 04/10/2018 02:26 PM, Ralph Böhme wrote:
> Hi Chris,
> 
> On Tue, Apr 10, 2018 at 12:30:42PM -0500, Christopher R. Hertel wrote:
>> I don't know if it is of any use to you at this point but, as mentioned
>> earlier, Albert (VMware) and I managed to put together a small number of
>> patches that properly handle the protocol piece of this.  Our code passes
>> the smbtorture and Windows Protocol Test Suite tests for Persistent Handles,
>> but doesn't have the backing store you are building into dbwrap.
> 
> So the Windows test suite doesn't really test server node failures? Hm.

The test suite would need to be able to tell one of the server nodes to shut
down (which, I suppose, it can do with the right controls).

>> Perhaps you've already referenced it.  It was based on work you and Michael
>> Adam had done previously.  It's in the repository I set up a while ago:
>> https://git.samba.org/crh/ContinuousAvail.git/
> 
> glad to hear you made progress and got something working.
> 
> Well, I guess my patches cover your basic option and SMB layer processing one
> way or another.

Worth a look, if you get a chance, just to make sure all the t's are dotted
and the i's are crossed.

> The trickiest part is getting the dbwrap backends and API right. Once that is in
> place, the rest is just plumbing it through the layers. :)

Agreed.

When I tried simply turning on persistent mode for the key databases, what I
got for my trouble was core dumps.  That was my first clue that the problem
would be more complicated.

Chris -)-----