Swarm Coordinator is not coordinating

This might be faster over IRC. Can you go to http://webchat.gamesurge.net and do: /j #Team-Rocket jumpboots

I’m currently in there as ‘j3rky’.

Yes.

charlimit

we resolved this over IRC. The problem was that the AllowedRemoteAgentGroup and AgentGroupName were not the same, so the two agents would think that they were part of different groups and reject each other’s jobs.

Gerke found out the problem.

AllowedRemoteAgentGroup and AgentGroupName should have the same string (I put “asd” on both).

Thanks Gerke!

Hi,
I have the same issue and configuration, but putting the same AllowedRemoteAgentGroup and AgentGroupName didn’t change anything.

Also building a (huge enough) scene gives me

Firewalls are disabled, pcs are on the same network, working fine with v-ray spawner for 3dsmax with a similar configuration.

Any idea ?

I just noticed that pinging coordinator during a rendering fails, while pinging coordinator when no rendering is on works fine.

I tried to put the project folder in a shared network folder to be sure that both agents can access (and even write) the folder, but it doesn’t work either.

In my case, installing DotNet Framework 4.6 fixed the problem.

For me - Allowing file and printer sharing in the windows firewall got it working. Apparently file and printer sharing also allows/disallows ping for win 8 (at least).

Thanks, that fixed the problem for me on windows 10

Although quite old thread, my experience with getting it to work (2 Windows 10 Pro systems) today after struggling 2 days:

For me, the problem did get solved by copying over the complete DotNET folder from UE4 to the slave machine (not just the 6 agent and 2 coordinator exe/dll files as shown in https://docs.unrealengine.com/en-us/Engine/Rendering/LightingAndShadows/Lightmass/Unreal-Swarm-Overview)

And after all my tests, I can confirm, that no shared directories are needed, as found in many discussions on this topic.

Here’s my working agent config (testing with non-defaults) - same on both systems, except for on the second system, I

  • reduced RemoteJobProcessorCount to 7,
    as load makes Windows unresonsive
    otherwise
  • chose some other CacheFolder directory

And developer settings

Important: For Local Rendering (SwarmCoordinator not available), set EnableStandAloneMode to True to avoid delays produced by first trying to contact the coordinator.

Basic setup done as described in UE4 Swarm Agent network distribution - YouTube

I did try many other things before which did not help at all, e.g.

  • Configuring names instead of ip addresses
  • Disabling firewall just in case
  • Checked Dot NET Framework install status
  • Checked UE4PrereqSetup.exe install status
  • Played with many config options

Nothing helped

I hope, this might be helpful for others, running into the same issues.

Hi!
I just change AllowedRemoteAgentNames to * and it works.

Thank you so much! I’ve struggled so much with this, for a good couple of hours, searching answers everywhere, but this is the only thing that helped. Old thread but only good answer for me. My hero for today herb :slight_smile:

Did you do this on all machines or just one hosting the project?

Just adding what combination of things worked for me for posterity, as I arrived here via search:

For reference I’m using two computers on the same LAN; one running the coordinator and agent tools, as well as the UE editor to kick off the lighting build, and a second machine just running the swarm agent.

Settings:

  • AgentGroupName and AllowedRemoteAgentGroup both set to Default on both machines.
  • CoordinatorRemotingHost set to 127.0.0.1 on coordinator machine, and hostname of the coordinator machine on the other machine.
  • AllowedRemoteAgentNames set to * on both machines.
  • Both machines can ping each other using regular Windows ping and get a response.
  • Both machines can use the in-app coordinator ping and remote agent pings to get responses from each other.
  • Both machines show up in the coordinator’s machine list.
  • The final thing I had to do, which I actually didn’t believe would help, was to make the project folder on the coordinator machine shared through regular Windows file sharing. Until I did this, jobs would just mysteriously not get picked up by the agent-only machine. I shared the parent folder of the project folder, if that matters. I’m also using the same Microsoft account user on both machines. I made sure I could access the folder on the agent-only machine through Windows Explorer first. I also did this trick to be able to actually log on with my MS account (MicrosoftAccount\mymsuser@microsoft.com).