On Mon, Dec 9, 2013 at 4:10 AM, John Arbash Meinel <[email protected]>wrote:
> > If (2) fails, we shouldn't have changed the machine; and if (3) > > fails I agree we need a fast path to just nuke the machine in > > state, but it's surely not going to leave the actual system in a > > confused state, is it? > > When I tried it, I ended up with > > "1": > agent-state: pending > ... > life: dying > > Which doesn't look like its ever going to resolve. > My contention is just that ending up in that state likely indicates a pre-existing problem with the implementation, that we shouldn't feel honour-bound to preserve -- but ofc the right response at this point may be to bug it and move on rather than jeopardize what you're working on. > Anyway, you sound very strongly in favor of moving all the code around > to support this, so I'll keep pushing on it. > Yeah -- while I *can* imagine cases where copy/paste is optimal given other constraints, I suspect this case is one in which the costs of that approach are excessive. Cheers William
-- Juju-dev mailing list [email protected] Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju-dev
