On Wed, May 11, 2011 at 12:02:18PM +1200, Tim Foster wrote: > Hi Krister, > > On Tue, 2011-05-10 at 16:10 -0700, [email protected] wrote: > > Folks, > > Here's a webrev for some zoneproxy fixes that didn't make the cutoff for > > b165. > > http://cr.opensolaris.org/~johansen/webrev-18277/ > > src/svc/svc-pkg-sysrepo, line 142 - nit: could this use "/usr/bin/pkill" > instead of "pkill"
Fixed. > src/zoneproxy/zoneproxyd/zoneproxyd.c > > Removing the option to pass configuration on the command line will make > testing a little more complex for this binary. Would it be worth > leaving those in, as an aid to test case development. If you think that you'll use this, then I'll add the code back in. > Also I wonder whether we'd ever want (or could support?) multiple > instances of the zone proxyd, with separate communication channels for > different zones, in which case, hardcoding a single system-repository > service instance seems wrong unless there are manual overrides available > from the CLI. Just echoing Ed's comments here, but it's probably better that we design for this case when we need to solve that set of problems. Running multiple zoneproxyd's shouldn't be a problem here, since they'd all be proxying to the same system repository destination. (Assuming that's ok...). Where this gets interesting is when you want multiple proxyd's mapping to multiple system repository instances. In that situation, we'll need some other bit of configuration information to tell us what system repository instance we're associated with. At that point, it would make sense to modify the config code to determine the instance to instance mapping, and read the right SMF properties. > Other than that, I'm not enough of a c-programmer to be able to properly > review the other changes, but I did look at them, and they seem ok. Thanks for taking the time to look at this. -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
