I’m experiencing issues, i have connected my Ferrum into 5-node cluster in which the main node functions only as an admin panel (gateway is closed), and other 4 nodes have the gateway active.
I can ping everything from the inside via wireguard IP’s, there is a passthrough between the nodes. All nodes have their gateway, except the fresh instance i just installed.
Of course i followed the Cluster guide (How to make a cluster - FerrumGate), and can confirm that Redis is also configured properly on the 5th node.
My desired implementation would be all child nodes connecting to main parent node to share config, with no passthrough between child nodes (it’s working as i expected), it’s just the one issue which i came across today - that no gateway appears even though it exists.
All of my ferrum nodes are in separate “regions”, and they communicate over WAN via the builtin wireguard.
Also a sidequestion - maybe You’d know, where could i look for - sometimes when connecting from just one network i am unable to estabilish connection to my peers and access services, but if i switch to wifi or mobile data instead it works perfectly, also in other networks it’s fine. There’s just that one network that sometimes allows to estabilish all tunnels, and sometimes it don’t.
I’d be grateful for help, FerrumGate is such great solution just what i was looking for, however the current issue just blocks my work.
Hi there,
We are working for a better multi region cluster configuration that is more easier than this one. And it will be ready to publish 1 month later.
If a gateway is not seen as not registered. It means that gateway could not reach redis on main node. On gateway machine try to reach main node redis over main 169.254.254.* ip.
Check also /etc/ferrumgate/env REDIS_PASS are same.
After gateway starting please check /etc/ferrumgate/env file REDIS_PROXY_HOST it will be like redis-ha:6379
sometimes when connecting from just one network i am unable to estabilish connection to my peers and access services
To solve this problem . on master machine follow logs with
ferrumgate --show-all-logs
Please feel free to ask any question. We are really appreciate for all feedbacks
Best regards
Yeah, i can confirm that i checked the .env files earlier, and it all looked correct, for a reference im attaching screenshots from env file from 3 nodes.
Here’s how i set a redis replica on new node (to replicate from master 169.254.254.193), You can see that it was earlier set properly too (and it pinged back):
The local redis password was different however i replaced it to check if that would do anything but of course it didn’t. Also i added the ES-HA just to make sure but yeah, it didn’t since it shouldn’t.
Here is a dump of docker processes from a new node:
And here is what i found on master node logs, it’s trying to sync the redis with new node and it looks like it did (i cannot see mentions about other nodes in the logs, only that one appears)
