upleb.uk

Public git repos — served from a NIP-34 GRASP relay at git.upleb.uk

summaryrefslogtreecommitdiff
path: root/src/net.h
diff options
context:
space:
mode:
authorBitflash <btf01@proton.me>2026-07-30 15:41:26 -0300
committerGitHub <noreply@github.com>2026-07-30 15:41:26 -0300
commit10f248876aaa77e840dcb0d9585aba2fde712130 (patch)
tree0591b8a78a47345e97720090c231b2a31f7bce49 /src/net.h
parent88a1e9973e34c972e965fe5dd16cd1f891885bba (diff)
net: notice a peer that has gone quiet, and stay audible to peers that check (#55)
A connection whose path dies silently -- NAT drops an idle mapping, a relay restarts, a route changes -- never delivers FIN. Nothing becomes readable, no error is raised, and the node simply stops hearing from that peer while the socket looks perfectly healthy. #51 bounded this for the rendezvous registration; the connection that comes out of the rendezvous had the same hole, and so did every ordinary peer. Stamp nLastRecv and nLastSend on the socket path and act on them: - nothing at all within 60s of connecting, disconnect. In testing this fired twice per run against dead .btf descriptors handed out by the relay, which is where the ghosts have been coming from. - queued to send for 10 minutes with nothing going out, disconnect. - nothing received for 30 minutes, disconnect. A healthy peer relays a block inventory about every two minutes, so this is fifteen intervals of silence. Bitcoin used ninety minutes against ten-minute blocks. Disconnecting is the whole cure: reconnection already works, and a peer we cannot hear is worth what no peer is worth. A node with nothing to relay would now look dead to a peer applying the same rule, so send "ping" after ten minutes of having said nothing. It carries no payload and expects no answer -- arriving is the entire content, because the receive path has already stamped nLastRecv by the time it is parsed. Requiring an answer would have read every node predating this change as dead, since those ignore unknown commands rather than replying. Verified on the live network: synced 0 to 3790 with five peers and no spurious disconnects, then again with the interval cut to 20s, which put 54 pings on the wire with no peer dropping us and no exceptions.
Diffstat (limited to 'src/net.h')
-rw-r--r--src/net.h32
1 files changed, 32 insertions, 0 deletions
diff --git a/src/net.h b/src/net.h
index 11c023e..449cff5 100644
--- a/src/net.h
+++ b/src/net.h
@@ -51,6 +51,24 @@ static const int64 PEX_MIN_INTERVAL = 60;
51// traffic is nothing: one per relay per five minutes. 51// traffic is nothing: one per relay per five minutes.
52static const int BTF_RENDEZVOUS_WAIT_SECS = 300; 52static const int BTF_RENDEZVOUS_WAIT_SECS = 300;
53 53
54// Liveness of an established peer connection. BTF_RENDEZVOUS_WAIT_SECS bounds
55// the wait for a dial to arrive; these bound the connection that comes out of
56// it, which has the same failure mode once it goes quiet.
57//
58// A healthy peer relays a block inventory roughly every nTargetSpacing (two
59// minutes), so silence measured in block intervals is the honest signal.
60// Bitcoin used ninety minutes against ten-minute blocks -- nine intervals --
61// and the same nine intervals here is eighteen minutes; thirty leaves margin
62// for a slow stretch without letting a dead path last an hour.
63static const int BTF_RECV_TIMEOUT_SECS = 30 * 60;
64static const int BTF_SEND_STALL_SECS = 10 * 60;
65static const int BTF_HANDSHAKE_GRACE_SECS = 60;
66
67// Send a ping after this long with nothing to say, so a peer running the
68// inactivity check above does not mistake a quiet node for a dead one. Must
69// stay well under BTF_RECV_TIMEOUT_SECS.
70static const int BTF_PING_INTERVAL_SECS = 10 * 60;
71
54// Descriptors to hand a peer: ours first, then peers that answered us. 72// Descriptors to hand a peer: ours first, then peers that answered us.
55void BtfPexCollect(std::vector<std::string>& vDescOut); 73void BtfPexCollect(std::vector<std::string>& vDescOut);
56// Verify descriptors a peer sent and remember the good ones. Returns how many. 74// Verify descriptors a peer sent and remember the good ones. Returns how many.
@@ -500,6 +518,16 @@ public:
500 // Last "btfpeers" we accepted from this node, to rate-limit the exchange. 518 // Last "btfpeers" we accepted from this node, to rate-limit the exchange.
501 int64 nLastPexRecv; 519 int64 nLastPexRecv;
502 520
521 // Liveness. A peer whose network path dies silently -- NAT drops an idle
522 // mapping, a relay restarts, a route changes -- never sends FIN, so the
523 // socket stays readable-never and the node simply stops hearing from it
524 // with no error anywhere. These stamps are the only way to tell that
525 // apart from a peer that merely has nothing to say.
526 int64 nTimeConnected;
527 int64 nLastSend;
528 int64 nLastRecv;
529 int64 nLastSendEmpty;
530
503 531
504 CNode(SOCKET hSocketIn, CAddress addrIn, bool fInboundIn=false) 532 CNode(SOCKET hSocketIn, CAddress addrIn, bool fInboundIn=false)
505 { 533 {
@@ -517,6 +545,10 @@ public:
517 nRefCount = 0; 545 nRefCount = 0;
518 nReleaseTime = 0; 546 nReleaseTime = 0;
519 nLastPexRecv = 0; 547 nLastPexRecv = 0;
548 nTimeConnected = GetTime();
549 nLastSend = 0;
550 nLastRecv = 0;
551 nLastSendEmpty = GetTime();
520 vfSubscribe.assign(256, false); 552 vfSubscribe.assign(256, false);
521 553
522 // Push a version message 554 // Push a version message