How to run >1 full node in a single network?

So i need to run 2 full nodes on the same network just because i have a shared network with someone else who is also running a full node of Chia.

I found that several people was able to run multiple full nodes on the same network, but i don’t know if its correct or not?

So basically need to disable UPNP both on the router and chia config level.
After that i need to change the 8444 port on the config as well.

Is this correct? thanks

1 Like

I’m only commenting because I’d like to know the answer to this one also! :nerd_face:

1 Like

There are 2 solutions to that.

  1. The same way as you can connect from many local computers to the same web server, chia can make outbound connections from many local clients. There are 2 lesser issues here. The first is that only 1 computer can port forward port 8444 to that particular box. From protocol point of view, there is no difference whether the connections are in- or out-bound, so just make sure that max 1 box has port forward (if any at all). The second problem is that all those boxes will start putting strain on the network, so depending on your ISP downstream bandwidth that may or may not be an issue (for normal operation bandwidth is in the range 2Mbps, so not a big deal, but it may go up during the syncing or dust storms).

  2. Chia has concept of privileged peers. What it means that such peer will be prioritized over all other peers. Therefore potentially a better solution would be to let only 1 box connect to the network and make the rest of local boxes act as slaves, and potentially also disable outbound connections. The main box will have to be considered trusted enough for those other boxes. So, for home setup (where the other box is just for a backup), this is a no brainer. If the other box belong to someone else, the other party has to trust that the main box will not try to make some unwanted things. There is no harm as far as security, the only thing that I can think of is that the main box may try to throttle those slaves.

You only should run UPNP on one node all others set to false. Don’t change the port and you shouldn’t need to change the router. I have done it for ages this way with no issues. The main node with UPNP enabled will have a lot of connections, the others will stop at 8. If they all on the same key use harvesters.

this port forwarding is from the router options?

im testing your suggestion, will come back again later.
last time i tried with the options i mentioned, the second full node can’t get the plots pass filter for pool points

i can’t get any pool points with this setting…
first full node has upnp enabled, second one disabled

Yes, this is done on the router. The end result (inbound connection) is the same as enabling UPnP on a box (as far as traffic). Maximum one box can have inbound connection.

You can disable UPnP on all boxes, and verify that router is not port forwarding. This will eliminate any issues with externally open ports. Both boxes should run fine with having just outbound connections.

By the way, pool communication has nothing to do with peers / UPnP / port forwarding. Pool communication is only one way from a box to the pool. So, it a box a box is not sending partials to the pool, there is potentially a problem with pool settings or outbound connections. To check the second one, you can check the peer section whether you have any (default outbound is 8 or so and is controlled via config.yaml).

My router setting with one UPnP already good i think

One node with UPnP enabled, the second one is disabled

the second node still have outbound conenctions but can’t get any partials to the pool

If you do see connected peers on the second node, then you may want to look into logs for pool related lines (debug.log). Usually, pools are good at providing errors if something is not going well. Of course, before pool replies, it first has to get partials, so you need to check whether partials are delivered to the farmer, and whether there are attempts to submit them.

Actually, just for testing, you can disable UPnP on your first box, and check whether the pool is still happy. If it is, then maybe the fastest way to check on the second box is to compare those logs right after node starts (on both boxes).