Lightmass swarm not working

Development Rendering
, , , ,
1

My swarm agents appear to be set up properly, but the remote agents never actually start computing lightmaps.

I get errors spammed to the local agent log:

[PushChannel] Pushing the channel has failed!

I’ve tried using wired connections to the same router as well as wifi connections, it doesn’t seem to make a difference. The graph shows the remote agent in a ‘connecting’ state for the duration. The remote agent’s log says:

[Connect] Remote agent connection confirmed: CHRIS-DEV-PC
[Job] Accepted job 57109.. etc

Here’s the local agent screens:

11293-capture.jpg
11293-capture.jpg1500×768 88.2 KB

11294-capture2.jpg
11294-capture2.jpg1500×768 173 KB

2

Here’s the coordinator machine:

11299-capture-coordinator.jpg
11299-capture-coordinator.jpg952×580 68.5 KB

3

Hi cgrebeld -

Thank you for your report. I will begin investigation into this issue as soon as possible. If possible could you please go to your CPU that is running the Swarm Coordinator and another just running the Swarm Agent and let me see the settings panel for each.

Thank You

Eric Ketchum

4

Here’s the agent-only machine:

11321-capture4.jpg
11321-capture4.jpg768×574 62.1 KB

5

Hi cgrebeld -

Try changing the Agent GroupName and AllowedRemoteAgentGroupNames to anything else and copy to both and try to rebuild and see if that will allow a better share of space.

Also it is best to use the CPU that is least powerful to be the Coordinator but start the build from one of the agents this will allow the coordinator to only function as the coordinator and distribute the processing.

Thank You and Let me know if either of those solutions work for you -

Eric Ketchum

6

Thanks for the reply. I tried both suggestions, but they seemed to have no effect. The job looks like it’s being distributed, but the remote agents never start calculation, and the “Pushing the channel has failed” error message is being spammed to the local agent.

7

Hey cgrebeld -

Instead of entering the name of your CPU in CoordinatorRemoteHost, look up and type in the IP address, it will probably be some sort of local IP address like 192.168.0.5.

Eric Ketchum

8

Hi,
I got it working by following one of the answers here:

I had to copy the files from the 7z file from dropbox into my local installs! I guess the missing files is due to a bug in your distribution packaging process.

I’m still on 4.2.1, so I’ll let you know if 4.3 requires the same hacks.

9

hello, i tried with 4.3.1 and followed all the steps (including adding the dropbox files and the patch files), but it’s not working. would like confirmation from epic that swarm is indeed working at least on their end.

i am on a gigabit LAN and have been able ping the coordinator and agents successfully. but when i start the build process, only the coordinator machine is processing the build and all the agents remain idle.

10

Hi, I’m not Epic - but I can confirm that swarm works with 4.3.1 on our machines, once those binaries are added as described in the other bug: 3 x AutoReporter.* files go into DotNET and the .pdb file goes into Win64.

11

thanks for the reply. good to know it’s working. i read another thread about different .net install causing issues. will investigate that route and hope for the best.