Re: [Nbd] NBD wishlist items?

[Mike S. <sn...@gm...>]

On 6/19/07, Mike Snitzer <sn...@gm...> wrote:
> On 6/18/07, Wouter Verhelst <w...@ut...> wrote:
> > Hi!
> >
> > I've been working on NBD pretty much in a vacuum for most of the time,
> > with the only input from other people being mostly bugreports.
> >
> > I know you LTSP people use NBD in a number of cases, and that you guys
> > use some wrapper scripts. Which is cool and all, I guess, but I'd think
> > it'd be better if you could do whatever you need to do without having to
> > use wrapper scripts. Right?
> >
> > So, if there's any feature request you guys have, it'd be cool if you
> > could communicate them to me. No guarantee I'll implement them within
> > any defined amount of time (unless the request is accompanied by patches
> > ;-), but I do plan to eventually implement them.
>
>
> Wouter,
>
> I've got one I'd like to run by you: have the nbd-client detect that
> the kernel's nbd connection to the nbd-server is unresponsive.
>
> There is a window of time where a potentially long TCP timeout
> prevents the nbd (in kernel) from erroring out back to userspace
> (nbd-client).  But if the nbd-client could feel that the nbd isn't
> behaving the nbd-client could send a SIGKILL down to the kernel (nbd
> driver already aborts TCP transmit if a SIGKILL is received).

Thinking through this a bit further it feels like this monitoring
would be best left outside of the nbd-client because the associated
code _could_ undermine the integrity of the nbd-client itself.  Not to
mention it would pollute the nbd-client code.  If anything it should
be a separate 'nbd-monitor'.

Mike