"headless" interactive app?

Hi,

I have something which is pretty similar to an OOD Interactive app - it runs as a batch job under slurm and can be accessed by a web server, and there is a proxy server that provides SSL termination. The lifecycle of this job is managed by an API.

I would like to use OOD to take advantage of its existing proxying/PUN functionality (because the way I did it involves a lot of duct tape and baling wire). But I do not want a full interactive app in my case. I still want to be able to start/stop these jobs only from my API. And I don’t want this to show up in the list of interactive apps that a user can choose in the OOD UI.

Not sure if this question makes any sense.

To put it another way, I would like the following:

  • A batch job starts - we don’t need to care about how it starts, and it listens on some port (plain old HTTP) on a cluster node
  • OOD provides an HTTPS URL that proxies the host and port
  • Somehow I can retrieve that URL in order to provide it via my API.

So I guess I am asking if I can use the proxying/PUN functionality of OOD without using any of the other trappings of interactive apps.

Is that possible?

Doesn’t sound like it. We don’t currently support SSL backends, so getting the proxy to correctly proxy to an SSL origin would require some hack in the lua scripts.

Then beyond that, it sounds like you need all the functionality of an interactive app, without actually using the interactive app subsystem which suggests to me you’d have to (re)write all of that.

Thanks. The thing I am proxying would not be SSL, it would be plain old HTTP.

And I am not sure I need all the functionality, just the proxying.

I’ll have to look at how OOD does it, and maybe reinvent that…