[bug]: Multiple URIs when Internet Provider change IP address

[tribotecno]

Pre-Submission Checklist

• I have searched the existing issues and believe this is a new bug.
• I am not asking a question about how to use lnd, but reporting a bug (otherwise open a discussion).

LND Version

lncli version 0.21.1-beta commit=v0.21.1-beta

LND Configuration

[Application Options]
#reset-wallet-transactions=true
maxpendingchannels=30
#minchansize=500000
accept-keysend=true
accept-amp=true
alias=T4b4j4r4ORG
max-cltv-expiry= 4032
coop-close-target-confs=144
max-commit-fee-rate-anchors=100
#Conf da VIVO
#listen=0.0.0.0:9735
#externalhosts=vivojg.m3.srv.br:9735

Conf da PREDIAL

listen=0.0.0.0:9735
externalhosts=predialjg.m3.srv.br:9735
nat=false

[tor]
tor.active=true
tor.v3=true

deactivate streamisolation for hybrid-mode

tor.streamisolation=false

activate split connectivity

tor.skip-proxy-for-clearnet-targets=true

[routerrpc]
routerrpc.apriorihopprob=0.75
routerrpc.aprioriweight=0.25
routerrpc.attemptcost=10
routerrpc.attemptcostppm=1
routerrpc.maxmchistory=10000
routerrpc.minrtprob=0.001

Backend Version

Bitcoin Core 31.0.0

Backend Configuration

cat umbrel/app-data/bitcoin/data/bitcoin/umbrel-bitcoin.conf
onlynet=ipv4
onlynet=ipv6
onlynet=onion
onlynet=i2p
listen=1
listenonion=0
i2pacceptincoming=0
peerblockfilters=1
blockfilterindex=1
peerbloomfilters=1
bantime=86400
maxconnections=125
maxreceivebuffer=5000
maxsendbuffer=1000
peertimeout=60
timeout=5000
maxuploadtarget=0
dbcache=450
prune=0
txindex=1
datacarrier=1
datacarriersize=83
permitbaremultisig=1
maxmempool=300
mempoolexpiry=336
persistmempool=1
rest=1
rpcworkqueue=128
chain=main
onion=10.21.22.10:9050
i2psam=10.21.22.11:7656
minrelaytxfee=0.00000100
blockmintxfee=0.00000001
incrementalrelayfee=0.00000100
rpcallowip=10.21.0.0/16
rpcallowip=127.0.0.1
zmqpubrawblock=tcp://0.0.0.0:28332
zmqpubrawtx=tcp://0.0.0.0:28333
zmqpubhashblock=tcp://0.0.0.0:28334
zmqpubsequence=tcp://0.0.0.0:28335
zmqpubhashtx=tcp://0.0.0.0:28336

[main]
port=8333
bind=0.0.0.0:8333
whitebind=0.0.0.0:8335
bind=10.21.21.8:8334=onion
rpcport=8332
rpcbind=10.21.21.8
rpcbind=127.0.0.1

OS/Distribution

Ubuntu 24.04.4 LTS

Bug Details & Steps to Reproduce

On my setup running in hybrid mode, changes to the ISP IP address are accumulating. Look at the lndgetinfo output:
],
"uris": [
"024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.5.85:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@oruqvgm6iaajssvkdzviuexfxudikpv5xosuwox5t34xvz46jfiv3gqd.onion:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.59.87:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@189.1.138.91:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.6.141:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@189.1.138.200:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.7.82:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.136.157.155:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.5.60:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.7.112:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.60.22:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.7.178:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@177.12.60.33:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@189.1.135.123:9735"
],

Only the last entry is valid. The configuration I use is below:
[Application Options]
#reset-wallet-transactions=true
maxpendingchannels=30
#minchansize=500000
accept-keysend=true
accept-amp=true
alias=T4b4j4r4ORG
max-cltv-expiry= 4032
coop-close-target-confs=144
max-commit-fee-rate-anchors=100

PREDIAL Conf

listen=0.0.0.0:9735
externalhosts=predialjg.mmm3.srv.br:9735
nat=false

[tor]
tor.active=true
tor.v3=true

deactivate stream isolation for hybrid-mode

tor.streamisolation=false

activate split connectivity

tor.skip-proxy-for-clearnet-targets=true

Any tips to fix this?

Expected Behavior

"uris": [ "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@oruqvgm6iaajssvkdzviuexfxudikpv5xosuwox5t34xvz46jfiv3gqd.onion:9735", "024a7699dd6d5159a1893050ef9f21d2f5ff6902f1e470c94ca282b6182b990ef1@189.1.135.123:9735"
],

Debug Information

No response

Environment

No response

[github-actions]

Found 2 possible duplicate issues:

1. [bug]: URIS are not deleted when changed via the externalip config #10233
2. [bug]: lncli getinfo uris spammed by 77 pubkey@tor.onion entries #10328

If this issue is a duplicate, please close it and 👍 the existing issue instead.

🤖 Generated with Claude Code

[hieblmi]

@tribotecno As a temporary workaround could you try setting externalip=<current-ip>:9735 and restart. The startup should replace persisted addresses instead of reusing old ones.

Could you close this issue in favor of the flagged duplicates that are mentioned here?

[saubyk]

cc: @Abdulkbk wasn't this issue was fixed with? #10235

[ziggie1984]

PR #10235 fixed stale address retention for externalip, but issue #10952 uses externalhosts, which follows a different flow.

externalip is handled during startup. If it is configured, lnd now treats it as authoritative and does not merge old persisted node announcement addresses from disk.

externalhosts is handled later by HostAnnouncer. On startup, if no externalip is configured, lnd still reloads the previously persisted addresses from the source node. That can
include old IPs that were previous DNS resolutions of the configured hostname.

After startup, HostAnnouncer resolves the configured hostname. However, it only sees a flat list of already advertised addresses. It does not know which persisted IPs came from
previous resolutions of the hostname, so it cannot remove them. It can only remove an old hostname address if it observed the old-to-new DNS transition during the current lnd process
lifetime.

Example:

externalhosts=node.example.com:9735

Historical DNS results:

node.example.com -> A
node.example.com -> B
node.example.com -> C

Persisted node announcement before restart:

A
B

After restart:

1. lnd reloads A and B from disk.
2. HostAnnouncer resolves node.example.com to C.
3. Final announcement becomes A, B, C.

Expected result:

C

Actual result:

A, B, C

So #10235 did not fix #10952 because it only changed the externalip startup path. It did not make externalhosts participate in startup address reconciliation.

## Proposed Fix Direction

Treat externalhosts as a configured address source during startup, not only as a later runtime updater.

At startup, lnd should build the desired clearnet address set from:

configured externalip addresses
+ currently resolved externalhosts addresses
+ NAT-discovered addresses, if NAT is active

Then it should preserve unrelated non-clearnet addresses, such as existing valid onion/I2P addresses, but not carry forward stale persisted clearnet IPs that are no longer part of the
configured/current address sources.

If both externalip and externalhosts are configured and produce different addresses, lnd should advertise both, since the operator explicitly configured both sources. If they produce
the same address, lnd should dedupe it.

The fix should also avoid pruning persisted clearnet addresses if startup DNS resolution for externalhosts fails, because otherwise a temporary DNS failure could remove the node’s only
clearnet address.

[Abdulkbk]

#10235 fixed a different issue as @ziggie1984 stated.

I can work on a fix if not assigned yet.

[saubyk]

@Abdulkbk assigned to you