Patch: Implement FSCTL_LMR_REQUEST_RESILIENCY
obnox at samba.org
Fri Oct 10 02:37:35 MDT 2014
On 2014-10-09 at 16:43 -0700, Jeremy Allison wrote:
> On Fri, Oct 10, 2014 at 12:07:14AM +0200, Stefan (metze) Metzmacher wrote:
> > Am 09.10.2014 um 23:58 schrieb Michael Adam:
> > > On 2014-10-09 at 14:45 -0700, Richard Sharpe wrote:
> > >>
> > >> While we are arguing about all the abstruse technical details, the
> > >> people I did this work for have reported that this change, along with
> > >> one other thing, seems to be enough to have HyperV create a Virtual
> > >> Disk on a Samba server, start the VM and install Windows.
> > >>
> > >> I hope to be able to write up the steps involved soon.
> > >
> > > That is great to hear!
> > >
> > > But these details are not so abstruse at all I'd say,
> > > but essential to our implementation of these features.
> > > After all, if we offer the possiblity to run vm images
> > > off a samba server, we have to make it vaguely safe I think.
> > We already have a "fake oplocks" option, we could also have a "fake
> > resilient handles" option...
> Perfect :-). Of course we should deprecate it as soon
> as the real underlying code is working :-).
Yeah, we can do that.
But we have to make sure that the documentation
carries a big fat disclaimer. Also the code
should spit out warning debug messages when these
We could do the same for persistent handles,
ca shares, etc. We have code that implements
parameters to trigger these options since >= 2 years
but only used them for test and development purposes.
We never proposed them upstream, because there is no
flesh behind them yet, and they are hence dangerous
if used in real setups. But I could imagine bringing
these to master with the "fake " prefix as tracer
ammunition so that people can see in test setups where
the journey is leading...
I will prepare some patches...
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 198 bytes
Desc: Digital signature
More information about the samba-technical