Connecting directly to the cluster peers

[Luflosi]

/ipns/cluster.pkg.pacman.store contains five multiaddrs. I tried to see if I could connect to the three IPv4 ones directly using ipfs swarm connect but I always get errors like "Error: connect 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx failure: context deadline exceeded".
I tried connecting from two different servers in two different locations.
Is this supposed to work?

[RubenKelevra]

Hey, sorry for the delay, the notifications for discussions were off.

We talked about this before, it was a remanence of an older iteration of this project.

I fixed it in 09b33cc, so I think this can be "closed"? :)

[Luflosi]

It still doesn't work as I expect it to. Executing ipfs swarm connect to any of the IPv4 peer addresses listed in service.json doesn't work.

$ ipfs swarm connect /ip4/45.83.104.156/tcp/16587/p2p/12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx
$ ipfs swarm connect /ip4/94.176.233.122/tcp/16587/p2p/12D3KooWGCSifNrJPZPfEdhAjRtxCW2dukiqQEqg4RAV6hE2jgbA

prints one of the following (output for the second command omitted as it is basically the same):

Error: connect 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx failure: failed to dial 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx:
  * [/ip4/45.83.104.156/tcp/16587] dial tcp4 0.0.0.0:4001->45.83.104.156:16587: i/o timeout

Error: connect 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx failure: context deadline exceeded

Error: connect 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx failure: failed to dial 12D3KooWDM4BGmkaxhLtEFbQJekdBHtWHo3ELUL4HE9f4DdNbGZx:
  * [/ip4/45.83.104.156/tcp/16587] failed to negotiate security protocol: message did not have trailing newline

and

$ ipfs swarm connect /ip4/140.238.249.171/tcp/16587/p2p/12D3KooWN9pSnzmE57cN4XDoYPRs1htjSRF2ig528kwb3VRrvf1Z

prints

Error: connect 12D3KooWN9pSnzmE57cN4XDoYPRs1htjSRF2ig528kwb3VRrvf1Z failure: failed to dial 12D3KooWN9pSnzmE57cN4XDoYPRs1htjSRF2ig528kwb3VRrvf1Z:
  * [/ip4/140.238.249.171/tcp/16587] dial tcp4 140.238.249.171:16587: connect: no route to host

and

$ ipfs swarm connect /p2p/12D3KooWL3V95pdTwumVCqbxWzuyCud3DqawJBi7koGos1iLwESQ

prints

Error: connect 12D3KooWL3V95pdTwumVCqbxWzuyCud3DqawJBi7koGos1iLwESQ failure: routing: not found

whereas connecting to my own peer from a second peer works:

$ ipfs swarm connect /dns4/luflosi.de/udp/4001/quic/p2p/12D3KooWBqQrnTqx9Wp89p2bD1hrwmXYJQ5x1fDfigRCfZJGKQfr

connect 12D3KooWBqQrnTqx9Wp89p2bD1hrwmXYJQ5x1fDfigRCfZJGKQfr success

I executed every command except the last one on two machines in two different networks.
I also find it a bit odd, that the list of trusted peers contains peer 12D3KooWL3V95pdTwumVCqbxWzuyCud3DqawJBi7koGos1iLwESQ, which is not listed anywhere else. Peer 12D3KooWN9pSnzmE57cN4XDoYPRs1htjSRF2ig528kwb3VRrvf1Z is listed in service.json but not in the README. The peer run by @teknomunk starts with 12D3KooWQkhWQYp according to the git history of the README but is not listed in service.json.

[RubenKelevra]

Well, that's expected. IPFS cannot connect to the IPFS-Cluster ports. :)

[RubenKelevra]

@Luflosi is your question now answered? :)

[RubenKelevra]

Closed and locked due to inactivity.

You can always connect to the servers in the list published in the README via:

$ ipfs swarm connect /ipv4/$ip/443/$ipfs_id

or

$ ipfs swarm connect /ipv6/$ip/443/$ipfs_id

The port you used before is the port for the cluster daemon connections, which isn't going to work with IPFS itself. Same goes for the IDs. This are cluster member IDs.

Apart from that, this shouldn't really be necessary, as the cluster members are all running gossipsub as pubsub router. So they should resolve extremely quickly to a small list of nodes which take part in the IPNS.