If a task is in progress on a client, at present you can't start another one. That's mainly a consequence of some long-standing limitations in Ghost's networking code - the GhostCast software is written to require everything to happen in very precise lock-step - that have ended up dominating how folks think about what tasks to, to the extent that the limitation has been carried right through into the console UI. That UI is essentially designed first and foremost to "script ghost" and all the other things that can be done in tasks end up being secondary to that.
[ The question of just how people use GSS is one that gets a lot of debate here; since we don't have any way at present of gathering usage statistics, and we're not sure how many people would opt in if we did, it's hard to know just what percentage of GSS customers use the console versus manual Ghosting and for console users just what the mix of tasks they run is. It's something I'd love to know more about, but the nature of things is that it's hard to get data on, especially to weight any input properly over the whole customer base (bigger customers have better access to product managers, f'rinstance). ]
I'd definitely like to remove this limitation, since it'd make the kinds of tasks that don't involve Ghosting over the machine a lot more useful. However, to really do that we need to transform the way tasks work internally pretty radically - the existing representation of tasks (including the console notion of "generic task" ) isn't suited to letting them queue up.
By the way, a closely related problem to queueing up task execution that I also want to solve is having a client machine manageable from more than one server. This would be great both for failover purposes, and because there are different kinds of management operations on machines (e.g. provisioning machines, versus provisioning applications, versus various kinds of routine helpdesk or other tasks).
Those things are just what I'd like to do, though - being merely a developer, my opinion doesn't make budget appear or convince product management to put features onto official product plans. If you want to see GSS change, the more people who ask for something the more data we can put in front of management to get the work funded.