render node licensing w/3rd party managers

Hi - just looking for clarity on what's necessary to use the "unlimited render nodes" for Fusion along with a renderfarm. Right now we're on Deadline, but will probably move to OpenCue soon, I think.

Since our farm is cloud based, the specific question I have, is does the rendernode require a license?

In the manual (pg 95) it says:
License dongle, Render Master, and Render Nodes must be on the same local network (subnet)

so how does that apply to a cloud based setup using a render manager like Deadline? Likely the machine with the now serial activation scheme wouldn't be on the same subnet as the render machines. Possibly but I'm not sure - they will have access to the same file system for sure, but may not be able to talk to each other really as far as ports and other usual "subnet" communication. Which is not an issue with a 3rd party manager, but would nix the built in one and likely wouldn't be able to sense a dongle without some fancy sort of vpn sharing with a local machine or something like that

Does the Render Node still perform this sort of licensing check, or does it somehow read the project to make sure it was created with the Studio version before rendering (but otherwise not need any special licensing)?

Any more info here would be great - I don't have a way to test it without going down a deeper road than I'm wanting to, so hopefully someone knows.

cheers!

My understanding from the manaul and some basic testing of my own is that the Render Node needs network access to a machine that's running Studio - or more specifically, running the "Fusion Server" process which is installed alongside Studio.

When launching a Render Node it scans the local LAN to find such a system.

You can use the environment variable FUSION_LICENSE_SERVER to point to a specific IP of a system running Fusion Server - so that could be on a different LAN, as long as it's network connectible on certain ports:
TCP port 1144
UDP port 1144
TCP ports 49152-49159
UDP ports 49152-49159

This environment variable can be set automatically in the Fusion Studio config file.

The Fusion Studio instance can be licensed either using a USB dongle, or (if using Fusion Studio 17) an activation key. Either is fine for licensing unlimited render nodes. I guess in a cloud setup, activation key would be preferable.

Check page 85+ of the current Fusion Studio 17 manual for more info.

Thanks Tom - hit the nail on the head, exactly what I needed. I think we can make that work - seems we'd just have to "burn" a license of Fusion Studio so that process is running (and reachable) on our cloud license server, that the renderfarm sees.

Appreciate the help!

seems we'd just have to "burn" a license of Fusion Studio so that process is running (and reachable) on our cloud license server, that the renderfarm sees.

If you use dongles, then yes probably it'd need to be dedicated to the farm. Technically you could have a workstation that was accessible from the cloud (through a VPN connection perhaps), and someone could use Fusion Studio on that workstation at the same time. But you'd have to ensure it never got rebooted else that would interrupt the farm.

Alternatively, an activation key license can run on two machines simultaneously. So if you have such a license, you could set up a Fusion Studio instance in the cloud (in a VM maybe) using that activation key license, and the same key could also be used on a separate machine, as a normal Fusion Studio workstation instance. So in that scenario you wouldn't have to dedicate a license to the render farm, only half of it.

On the subject of Fusion render managers, Andrew Hazelden recently wrote a great post over on WSL describing using Houdini for automating Fusion rendering with logic. UserNoah then expanded on the idea and has made a working (but still considered alpha) render manager.

https://www.steakunderwater.com/wesuckl ... php?t=5165

Could be very interesting for anyone who needs more advanced logic in their render managing.

Saw that! Looked like there's a lot of potential there and I love being able to get at the bits and chain up the dependencies with a graphical interface. That could be super cool.
I like the idea of being able to possibly distribute Resolve renders, too. I haven't bumped into the need for it yet, but I could imagine a Resolve session could get pretty heavy in the longform world with feature work. And also, as the Fusion integration in Resolve continues to get better, eventually I could see attempting that unicorn of doing the comps in Resolve (or at least transferring from the standalone to there) and having full blown comps living live, right in the edit. Being able to distribute that sort of thing to a few machines come render time could be alriiiiiiight.