topology crash on update to v25.02

[michael1011]

After updating to v25.02, the node crashed with:

2025-03-11T18:54:52.181Z **BROKEN** plugin-topology: Plugin marked as important, shutting down lightningd!
2025-03-11T18:54:52.181Z INFO    plugin-topology: Killing plugin: exited during normal operation




0x0 ???
  ???:0
0x7f48d53da51f ???
  common/daemon.c:78
0x5646625c2b49 crashdump
  common/daemon.c:33
0x5646625c2aad send_backtrace
topology: FATAL SIGNAL 11 (version v25.02)

I removed the gossip store and restarted the node, ever since then, my channels are not becoming active and I have no active peers
The startup issue was caused by DNS

[kilrau]

Linking Rusty's DNS issue #8157

[gallizoltan]

I have the same issue. The first time, lightningd ran for hours before crashing:

2025-03-14T03:32:58.279Z INFO    plugin-keysend: Inserting a new invoice for keysend with payment_hash e23a23e63d0ced24949f4d59e9cf370c3a30e36671afffcca0cadbac443a6294
topology: FATAL SIGNAL 11 (version v25.02)
0x55af18c1aaad send_backtrace
        common/daemon.c:33
0x55af18c1ab49 crashdump
        common/daemon.c:78
0x7f38a6d2951f ???
        ./signal/../sysdeps/unix/sysv/linux/x86_64/libc_sigaction.c:0
0x0 ???
        ???:0
2025-03-14T03:33:02.395Z INFO    plugin-topology: Killing plugin: exited during normal operation
Missing result from listincoming: {"jsonrpc": "2.0","id": "cln:htlc_accepted#1678463/keysend:invoice#3/cln:listincoming#1678468","error": {"code":-4, "message":"Plugin terminated before replying to RPC call."}}
2025-03-14T03:33:02.395Z **BROKEN** lightningd: Missing result from listincoming: {"jsonrpc": "2.0","id": "cln:htlc_accepted#1678463/keysend:invoice#3/cln:listincoming#1678468","error": {"code":-4, "message":"Plugin terminated before replying to RPC call."}}
lightningd: FATAL SIGNAL 6 (version v25.02)

After the first crash, I cannot even start lightningd, it crashes instantly:

2025-03-14T03:46:36.447Z INFO    lightningd: Server started with public key 03d67f36c4f81789e2fe425028bacc96b199813eae426c517f589a45f1136c1fe5, alias Jubilee (color #dc42f4) and lightningd v25.02
topology: FATAL SIGNAL 11 (version v25.02)
0x560037f64aad send_backtrace
        common/daemon.c:33
0x560037f64b49 crashdump
        common/daemon.c:78
0x7f6c41ff351f ???
        ./signal/../sysdeps/unix/sysv/linux/x86_64/libc_sigaction.c:0
0x0 ???
        ???:0

[rustyrussell]

Please send me your gossip_store file so I can reproduce?

[gallizoltan]

I tried, but I couldn't attach it to this ticket. Could you please download it from here?
https://louismartin.hu/gossip_store
I modified the chmod attributes to make it accessible on this site. If it matters, you can change it back with chmod 600 gossip_store.

[rustyrussell]

I cannot reproduce this with that gossip_store, at all :(

I'm frustrated that we're not getting decent backtraces, either. I assume you've deleted the gossip store and it no longer happens.

Two options if it happens again:

1. Try a debug build (./configure --enable-debugbuild).
2. Try make clean and rebuild in case there's some mismatch?

[rustyrussell]

Nope, found it. The clue was to note that one of your incoming channels has an unrepresentable fee level, and we tried to log it, only the log function for that path is NULL.