Deadline RCS Server. TBs of Network Traffic!

Deadline
1

Hi all,

Running Deadline 10.2.0.10 under AWS. The Server is in RCS mode. We have a max of 30 Artists with Deadline Monitor open. And ~40 Spot Instances up and down throughout the day.

Last month, we ended up with some crazy bandwidth under an account. So I began troubleshooting.
Over the last 3 days, 1.5TB of Network Traffic has occurred on the Deadline Server. (mostly OUT)

I have no hard data isolating Artist Deadline Monitor connections vs Deadline Worker (Spot Instances), but it seems like ~45MBps per Deadline Monitor connection!

What would generate so much traffic from the Deadline Server to Deadline Monitor??? This is just so unexpected.

On the Deadline Worker side, the only Aux files we use are Nuke scene files… I checked the repo and there is a total of 2.3GB of .nk files… so let’s say all 40 nodes picked up for all those jobs… still just 92GB)

Thanks for any insights!

-Paul

1 Like
2

Heya, just wondering if you ever managed to track this down? We have an issue at the moment where the RCS’s is bottlenecking, maxing out it’s 24GB of RAM and causing the monitor to become unresponsive. I am wondering if it is related somewhat with the server storing all these network commands in it’s RAM.

For context we have around 300 local workers and 60 so artists and are trying to find the root cause of this.

3

Are asset files also transferred? e.g. images, geometry. This is the stuff that usually takes up space.
Do you see anything interesting in the logs? Both the Deadline RCS, and the AWS Asset server, etc.
If too many assets are being uploaded at once, I can imagine this depleting some resources.