AWS Thinkbox Discussion Forums

How to change port forwarding to several rendernodes behind a NAT router?

hey there,

we´re moving from on site office to remote office only and I have a few issues figuring out how to make this work.
We´re going to use locally network mapped drives with the same drive letter on all remote office workstations as well as on our large office QNAP NAS and then sync them all via a cloud storage provider.
We´re going to have workstations+rendernodes at each remote office and some rendernodes attached in a lokal network where the QNAP NAS stands.
All rendernodes and workstations are behind NAT routers, so we´re going to give each of them a separate DynDNS.
Using portforwarding In the office router I could set up so I can now control the workstation in the office running the Deadline Connection Server/MongoDB.
But for all other rendernodes I am at a loss on how to:
A) power them on and off
B) Start workers as needed.

I found a tutorial on how to set up portforwarding for several PCs behind a NAT, using differen external and lokal port combinations:

Unfortunately our router doesn´t allow that (it keeps telling me a port is already in use, if I´m trying to forward the same external port to a different lokal port on another machine.).

I´ve tried adding the specific machines DynDNS in its worker properties, as well as changing the default por (5054) to something else, and then forwarding that port in the Router to that machine.
But I still cannot remote control it.

I just tried to ping the second workstation from a remote location via the DynDNS specified in the Worker properties and the ping timed out.
Pinging it via the same network from the first workstation withthe DynDNS hostname worked though.

Any Pointers would be greatly appreciated!

I can’t say anything worth hearing on the networking side of things, but I can warn you that Deadline’s power management uses wake-on-lan ‘magic packets’, which can’t really leave the LAN they’re created it.

So you may benefit from setting up Deadline regions for each LAN and running a Pulse in each LAN for each region. So that way each Pulse will be generating WOL packets for each LAN. That might be overcomplicating things though, so I’d love to be corrected. :smiley:

Ah, yes, but regions would probably make sense anyways, never had to use them before, but the way our farm will be spread around several remote offices, there might be benefits besides the WOL.
We´d still need to use a separate tool to wake up the remote workstations, unless we decide they´d be running nonstop anyways…

Hm, ok, I´m not sure I get how this works.
We could work around not being able to use WOL from outside the respective sites over internet, as long as the local rendernodes would be woken up through deadline.
So, how would we have to set this up?

  1. Have one workstation on each site run a separate pulse.
  2. Have each rendernode at each site assigned to a specific region via their launcher rightclick menu.

Can I now simply submit a job to the farm and all local pulses will wake up the rendernodes in their local LAN?

Yes, assuming you’ve got power management set up I’d expect each region’s Pulse to attempt to start the other Workers in that region.

Privacy | Site terms | Cookie preferences