Peer count stays 0

[AntoineXBTO]

System information

Geth version: 1.9.7-stable
OS & Version: Linux

• Issue similar to No IP address in enode addresses #2765
• I can't use UPnP so I need to add the external ip manually via --nat=extip:MYIP
• admin.nodeInfo gives me an enode that looks correct
• netstat shows geth listening on 30303 for ipv6 but not ipv4
• I cannot make any connection, inbound or outbound from that node, it has been running for hours without catching any peers

[AntoineXBTO]

I would like to add that admin.addPeer("...") to a valid node accessible on inbound doesn't log anything on either side.

[holiman]

I would suspect that it does, but on a pretty low level, like DEBUG or TRACE. Try temporarily setting debug.loglevel to 5

[AntoineXBTO]

I could finally connect to a second node. Yes loglevel to 5 could help me see the log.
The node doesn't seem to try to connect to any other peer though and only relies on that manually entered one.

[holiman]

Could you elaborate on your network setup? Also, does your netstat look like this:

tcp6       0      0 :::30303                :::*                    LISTEN      4595/build/bin/geth 
udp6       0      0 :::30303                :::*                                4595/build/bin/geth 

If so, it means both ipv4 and ipv6.
Is your external IP ipv4?

[fjl]

Please run geth with --vmodule "p2p/*=5" and post some logs here or in a gist. Thank you.

[AntoineXBTO]

@holiman it does look like this (notice the brackets)
tcp6 0 0 [::]:30303 [::]:* LISTEN

For security reasons I won't be able to describe the network setup too extensively but:

• there is not public ip on that server
• we're using a NAT with a public IP and I'm using that ip as parameter for geth --nat=extip:IP, yes it's ipv4
• firewall is heavily restricted, ports 30303 on TCP and UDP are open though, for outgoing connections

@fjl same here, I won't be able to export my logs but:

• it starts by successfully finding nodes in database
• starts p2p networking
• ping/v4 on those nodes, err=nil
• findnode/v4, err=nil
• findnode failed "RPC timeout" with failcount > 325

These nodes might be too old but what I want to understand is why it doesn't try to find new ones once connected to the node I manually connect to?
The connection to the manually entered node is "staticdial"

[fjl]

Connected peers do not influence peer discovery. How many other nodes does your node ping? Just one, or many?

[AFDudley]

I am having this same issue. using netstat -ntlp4 does not return 30303.

Running lsof | grep LISTEN returns something like:

sshd       5084          root    3u     IPv4              54687        0t0        TCP *:ssh (LISTEN)
sshd       5084          root    4u     IPv6              54689        0t0        TCP *:ssh (LISTEN)
geth      26619 38003    geth  165u     IPv6         2655098132        0t0        TCP *:30303 (LISTEN)
geth      26619 38003    geth  174u     IPv4         2656677356        0t0        TCP 127.0.0.1:8546 (LISTEN)
geth      26619 38003    geth  179u     IPv4         2656677355        0t0        TCP 127.0.0.1:8545 (LISTEN)

Note that it states when explicitly binding to both addresses.

[holiman]

cc @fjl ^

[fjl]

Geth binds to 'any address', which covers both IPv4 and IPv6. Command line tools usually show that as IPv6 because it's a single listener. Your sshd example looks different because it uses two separate listeners for IPv4 and IPv6.

[dastageer-eth]

HI, @fjl i'm trying to run a execution client using geth, but it cannot find peers to start block download and initiate sync, what can I do?
I'm from India, if it has any relevance in finding peers