[SOLVED] Linux-based Cluster ID - How to make it work?

[LHammonds]

  

 

I am setting up a headless Ubuntu 20.04 Server to host multiple Ark maps and have it working but not as a cluster yet.

I already have this folder setup (with correct permissions):

/opt/ark/cluster

I start both maps with the following parameters in the startup arguments (reduced to just options related to clustering and fake ID):

-clusterid=1122334455 -ClusterDirOverride=/opt/ark/cluster -NoTransferFromFiltering

I can see that once the maps are online, they create the following folder:

/opt/ark/cluster/clusters/1122334455/

But there are no files inside that folder and when I access an obelisk, it does not show the other map regardless of the checkboxes selected.  I can join either one directly but they are not visible inside the in-game transfer.

Is this because it has to have Steam-specific ports open besides the ones specified for Ark usage (Query/Game/Raw ports?)   I have these ports in use and open on the OS and hardware firewall:
 

27051/udp - Island query port
7777/udp - Island game port
7778/udp - Island raw port
27065/udp - Ragnarok query port
7791/udp - Ragnarok game port
7792/udp - Ragnarok raw port

Thanks,
LHammonds

[LHammonds]

Found out what prevented this from working.  The server configuration was 100% correct.

The problem was use of 1-to-many NAT (Network Address Translation) on the firewall.  The firewall has many static IP addresses and apparently, the inbound external IP has to match the outbound IP and a 1-to-many NAT may have outbound traffic going out on a different external IP which is normally not an issue but Steam has an issue with it.

The solution was to change the firewall rules to a 1-to-1 NAT so that one external IP is dedicated to one internal IP (rather than sharing one external IP to many internal servers via different ports)

The hardware firewall used was a Meraki MX100

LHammonds