small legend:
B is subdomain like nodes
A is domain like mydomain.com
I am running an homelab and using a docker ddns client to keep “in sync” the IP for my nodes.
This ddns client has a subdomain like:
B.A
Each storj node I have has an ADDRESS env like:
B.A:portForEachStorjNode
so for example storj-node1 is exposed on
B.A:28967
while storj node 2 is exposed at
B.A:28968
and so on.
When I try to add one of those nodes to the multinode dashboard as
B.A:portOfAnode
I get like a connection time out.
Second problem when I reach subdomain.mydomain.com I reach the UI of my gateway (ubiquiti udm SE)
Any idea on how I can tackle this issue?
Explain where are you nodes located, network wise, and where are you trying to reach them from, what the addresses of nodes resolve to in your lan. You might need to adjust dns in your lan so that nodes resolve to lan address when queried from inside your lan.
My nodes are all running in the same server (I started as many do with a small SSD for experimenting and since it was full I started to use new disks).
These nodes are all in the same macvlan network (so each container has its own “physical” IP).
At the same time they are exposed on an “internal” docker network where multinode container runs, therefore storj nodes and multinode can reach each other.
To add one node to multinode, I reached the IP of myserver:15002 for multinode dashboard, I then added the external subdomain.domain.com:portOfNode apiKey etc but multinode in its logs shows a timeout.
In the UDM SE I configured the port forwarding to accept any traffic and based on the incoming port:
28967 redirects to 192.168.1.10 for node1
28968 redirects to 192.168.1.11 for node2
28969 redirects to 192.168.1.12 for node3
At the moment the UI of each node is reachable with traefik + pihole for local dns record like:
node1.home.A
When I try to reach the nodes through their subdomain.domain.com:portOfNode I don’t see any response in the browser
Yea, it’s a hairpin nat. You can configure it on UDMP, but it’s not advisable. Instead, add FQDN of your nodes to DNS resolver (even just on the instance that runs multinode) and point to their lan addresses. Even adding them to /etc/resolv.conf shall suffice. There is no need to reach the service via internet if you are in the same LAN. Or even just use lan addresses to add nodes.
Btw, when you reply to topic using the Reply To Topic button,
the person you are actually replying to, or anybody else in the topic, for that matter, is not notified. So you end up waiting for response much longer, until that person stumbles on the same topic again as a result of sheer boredom. Instead, use a reply button:
@andrew2.hart@arrogantrabbit
Am I an idiot
I run all the storj nodes in the same docker-compose along with multinode and I forgot to add the networks section to multinode, therefore multinode was on its own isolated network which is different from the network where the nodes were.
Cool to notice that the dashboard can be used fully locally, and I don’t need to add the external IP of the nodes!
Somehow I just gave for granted the multinode container MUST use the public IPs of the nodes and not the internal IP or even better, FQDN.
