Alan Coopersmith wrote:
> Garrett D'Amore wrote:
>
>> Actually, I think what would be nice here would be some form of UNIX
>> domain socket or named pipes that crossed zone boundaries.
>>
>
> We already have those, as long as the endpoint is in a filesystem location
> visible to both zones. See for instance LSARC 2008/506, in which X11 clients
> in TX labeled zones connect to the Unix domain socket listener opened by the
> X server in the global zone. (There was a bug for quite a while preventing
> cross-zone communication across Unix domain sockets, but as discovered during
> LSARC 2008/506, that was a bug, since the original Zones case specified they'd
> just work [1].)
>
Cool. Shows how little I know about xVM zones.. :-)
So this is the kind of IPC that I guess these things should be using.
The only question that remains is locating a common directory where
software can rendezvous with these things. Sort of like /var/run, but
with a predictable location both in the global zone and in local zones.
( E.g. /var/run/zones/<zonename> in the global zone, and
/var/run/zones/self in the local zone, or something similar.)
I imagine that could be a fairly trivial fast track and RFE for someone
who really wants to implement it.
-- Garrett
> [1] Chapter 13 of PSARC 2002/174 Design Spec:
> http://opensolaris.org/os/community/arc/caselog/2002/174/zones-design.spec.opensolaris.pdf
>
>
