From a08444862b5492cec239cdca82f9cf32fa3da765 Mon Sep 17 00:00:00 2001
From: Tanmay Deep Sharma <tanmaydeepsharma21@gmail.com>
Date: Tue, 21 Apr 2026 18:12:32 +0700
Subject: [PATCH] fix(media-server): give inbound DTLS enough time for Meta to
 catch up
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Inbound calls consistently died at ~2s with the Meta PC going straight
from connecting → closed (never reaching connected). Root cause: pion
starts the DTLS handshake the moment we SetLocalDescription during
create_session, but Rails only delivers our SDP answer to Meta via
pre_accept_call / accept_call afterwards. During the ~1s gap Meta has
no fingerprint for us and drops our ClientHello. pion's default DTLS
retransmission window lapsed before Meta was ready to respond.

- Bump the DTLS connect context to a 30s timeout so retries keep going
  until Meta has our credentials.
- Shorten the retransmission interval to 200ms so a retry lands quickly
  once Meta is listening.
- Log DTLS transport state and the stack trace for explicit pc.Close()
  so the next regression is easy to diagnose.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 .../media-server/internal/peer/meta_peer.go   | 24 ++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/enterprise/media-server/internal/peer/meta_peer.go b/enterprise/media-server/internal/peer/meta_peer.go
index 7544b12da..5592ad4b3 100644
--- a/enterprise/media-server/internal/peer/meta_peer.go
+++ b/enterprise/media-server/internal/peer/meta_peer.go
@@ -3,9 +3,12 @@
 package peer
 
 import (
+	"context"
 	"fmt"
 	"log/slog"
+	"runtime/debug"
 	"sync"
+	"time"
 
 	"github.com/pion/interceptor"
 	"github.com/pion/webrtc/v4"
@@ -45,6 +48,17 @@ func NewMetaPeer(cfg *config.Config, sdpOffer string, iceServers []webrtc.ICESer
 		return nil, "", fmt.Errorf("set UDP port range: %w", err)
 	}
 
+	// Meta's side finishes setting up our DTLS credentials only *after* Rails
+	// calls pre_accept_call / accept_call on the Graph API. For inbound calls
+	// that happens ~1 second after we've already begun the DTLS handshake, so
+	// our first ClientHello is dropped. Extend the handshake window to 30s and
+	// tighten the retransmission interval so a later retry lands once Meta is
+	// ready.
+	se.SetDTLSConnectContextMaker(func() (context.Context, func()) {
+		return context.WithTimeout(context.Background(), 30*time.Second)
+	})
+	se.SetDTLSRetransmissionInterval(200 * time.Millisecond)
+
 	// If a public IP is configured, use NAT1To1 so ICE candidates advertise
 	// the correct address instead of a private Docker/container IP.
 	// NOTE: In pion/webrtc v4.2+, migrate to SetICEAddressRewriteRules.
@@ -154,6 +168,14 @@ func NewMetaPeer(cfg *config.Config, sdpOffer string, iceServers []webrtc.ICESer
 		slog.Debug("meta peer: signaling state changed", "state", state.String())
 	})
 
+	// Also log DTLS transport state — failure here is the likely cause of the
+	// inbound "closed immediately after ICE connected" symptom.
+	if t := pc.SCTP().Transport(); t != nil {
+		t.OnStateChange(func(state webrtc.DTLSTransportState) {
+			slog.Info("meta peer: DTLS state changed", "state", state.String())
+		})
+	}
+
 	// Perform SDP negotiation based on call direction.
 	var sdpResult string
 	if sdpOffer != "" {
@@ -261,6 +283,6 @@ func (mp *MetaPeer) Close() error {
 	}
 	mp.closed = true
 
-	slog.Info("meta peer: closing peer connection")
+	slog.Info("meta peer: closing peer connection (explicit)", "stack", string(debug.Stack()))
 	return mp.pc.Close()
 }