connected is not going to die when lightningd dies

[vincenzopalazzo]

Issue and Steps to Reproduce

When lightningd is dying for some reason, the other depending deamon is not going to die

So in my case is connected, so when I try to restart the node I get the following error

2024-11-21T10:57:36.725Z DEBUG   hsmd: new_client: 0
2024-11-21T10:57:36.794Z DEBUG   connectd: Proxy address: 127.0.0.1:9050
2024-11-21T10:57:36.794Z DEBUG   connectd: Failed to bind 2 socket: Address already in use
2024-11-21T10:57:36.794Z DEBUG   connectd: REPLY WIRE_CONNECTD_INIT_REPLY with 0 fds
2024-11-21T10:57:36.794Z **BROKEN** connectd: Failed to bind socket for 0.0.0.0:19735: Address already in use

getinfo output

running the commit a90d9c9

---

Not sure if this is going to help #7630

[rustyrussell]

Are you sure connectd is still running? This could be the TCP socket without SOREUSEADDR.

[vincenzopalazzo]

Are you sure connectd is still running? This could be the TCP socket without SOREUSEADDR.

not sure how to double-check this, but this is happening only with the current master, not with v24.08 I had the same issue in two difference machine

[rustyrussell]

ps auxwww | grep connectd ?

[whitslack]

I usually do a killall -u lightning after stopping my lightningd service because very often not all of the subdaemons terminate when lightningd does.

[kristapsk]

Not sure it is the same issue, but I'm having same error when trying to setup a new lightningd testnet instance on Ubuntu 24 VPS.

While testnet3 bitcoind is doing IBD.

neonz@bitcoin-testnet:~$ tail -F /data/lightningd/cln.log | grep -v DEBUG
2025-02-12T16:16:54.659Z INFO    lightningd: v24.11.1
2025-02-12T16:16:54.922Z INFO    plugin-clnrest: Killing plugin: disabled itself: No module named 'gevent'
2025-02-12T16:16:54.924Z INFO    plugin-wss-proxy: Killing plugin: disabled itself: No module named 'websockets'
2025-02-12T16:16:54.924Z UNUSUAL plugin-bookkeeper: topic 'utxo_deposit' is not a known notification topic
2025-02-12T16:16:54.924Z UNUSUAL plugin-bookkeeper: topic 'utxo_spend' is not a known notification topic
2025-02-12T16:16:55.032Z INFO    connectd: Static Tor service onion address: "yivhkj7z4oav2hk4uesjic62ldz6q4hj5lnhzsbhxknhp4nj222tqmqd.onion:9736,0.0.0.0:9736" bound from extern port 9736
2025-02-12T16:16:55.044Z INFO    plugin-bcli: bitcoin-cli initialized and connected to bitcoind.
2025-02-12T16:16:55.051Z UNUSUAL lightningd: Waiting for initial block download (this can take a while!)
2025-02-12T16:16:55.088Z **BROKEN** connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use

neonz@bitcoin-testnet:~$ grep 9736 /data/lightningd/cln.log | tail -n 4
2025-02-12T16:16:55.088Z **BROKEN** connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use
2025-02-12T16:17:55.502Z DEBUG   connectd: Created listener on 0.0.0.0:9736
2025-02-12T16:17:55.502Z INFO    connectd: Static Tor service onion address: "yivhkj7z4oav2hk4uesjic62ldz6q4hj5lnhzsbhxknhp4nj222tqmqd.onion:9736,0.0.0.0:9736" bound from extern port 9736
2025-02-12T16:17:55.551Z **BROKEN** connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use

neonz@bitcoin-testnet:~$ cat /data/lightningd/config
alias=rtbc.lv-test
#rgb=<your hex color>
network=testnet
log-file=/data/lightningd/cln.log
log-level=debug
# for admin to interact with lightning-cli
rpc-file-mode=0660
# default fees and channel min size
#fee-base=<1000>
#fee-per-satoshi=<1>
#min-capacity-sat=<your minchansize>
## optional
# wumbo channels
large-channels
# channel confirmations needed
funding-confirms=2
# wallet settings (replication recommended, adjust backup path)
wallet=sqlite3:///data/lightningd/testnet/lightningd.sqlite3:/home/lightningd/lightningd.sqlite3
# no replication:
#wallet=sqlite3:///data/lightningd/bitcoin/lightning.sqlite3
# network
proxy=127.0.0.1:9050
bind-addr=0.0.0.0:9736
addr=statictor:127.0.0.1:9051/torport=9736
always-use-proxy=true
# experimental options
# use anchor channels
experimental-anchors
# enable dual fund option
experimental-dual-fund

[ishi0k]

Hello everyone.

I got the same problem to updagre from v24.08 to v24.11.1

2025-02-24T16:29:07.166Z INFO plugin-bcli: bitcoin-cli initialized and connected to bitcoind.
2025-02-24T16:30:23.973Z INFO plugin-clnrest: REST server running at https://127.0.0.1:5010
2025-02-24T16:30:26.391Z BROKEN connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use

I deleted the gossip file, but I’m still getting this error: BROKEN connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use.

In the CHANGELOG v24.11.1, I see this:

Changed
Plugins: gRPC now starts on port 9736 by default (localhost, see grpc-host) (#7479).

And in my CLN configuration file, I have this:# network
proxy=127.0.0.1:9050
bind-addr=0.0.0.0:9736
addr=statictor:127.0.0.1:9051/torport=9736
always-use-proxy=true
Should I apply any changes to the configuration file for version v24.11.1?

[Schnema1]

Same here:

2025-03-28T21:58:00.828Z INFO lightningd: v25.02
2025-03-28T21:58:08.865Z INFO connectd: Static Tor service onion address: "i5***.onion:9736,0.0.0.0:9736" bound from extern port 9736
2025-03-28T21:58:08.914Z INFO plugin-bcli: bitcoin-cli initialized and connected to bitcoind.
2025-03-28T21:58:11.335Z INFO plugin-cln-grpc: Connecting to "lightning-rpc" and serving grpc on 127.0.0.1:9736
2025-03-28T21:58:11.675Z INFO plugin-summary.py: Plugin summary.py initialized
2025-03-28T21:58:11.782Z BROKEN connectd: Failed to listen on socket 0.0.0.0:9736: Address already in use

After downgrade to 24.11 and reboot of the system I get the same error.
But downgrade to 24.08 seems to be not supported.

2025-03-28T22:19:34.155Z INFO lightningd: v24.08.2
2025-03-28T22:19:35.076Z UNUSUAL plugin-bookkeeper: topic 'utxo_deposit' is not a known notification topic
2025-03-28T22:19:35.076Z UNUSUAL plugin-bookkeeper: topic 'utxo_spend' is not a known notification topic
2025-03-28T22:19:41.895Z BROKEN lightningd: Refusing to migrate down from version 261 to 256

So I am st(f)ucked.

[whitslack]

@Schnema1: You can't have CLN listening on the same port as its GRPC plugin. Maybe put the node back on the default port (9735).