Hello Blizzard Support,
I’m experiencing severe latency spikes (1,000–2,000 ms) in Diablo IV that appear to originate beyond my local network, likely on the path to—or within—the Google Cloud instance hosting the game.
What I measured
Platform / Region: PC (Windows), EU.
Server observed: 35.241.144.104:6112 (reverse DNS 104.144.241.35 bc google usercontent).
LAN health: Ping to gateway 192.168.1.1 stays <1 ms (no jitter/loss).
Route health: tracert 35.241.144.104 shows ~15–16 ms into Google; later hops don’t answer ICMP (expected), but Wireshark on the game flow shows heavy retransmissions and ACK RTT > 1s exactly when the lag occurs.
PC not congested: No tcp.analysis.zero_window events observed during spikes.
Evidence from the last PCAP (client port 64760 ↔ server 35.241.144.104:6112)
Wireshark flags (server → client) around the lag window:
78.746313 [TCP Previous segment not captured]
79.007963 [TCP Fast Retransmission] 6112 → 64760 Len=435
81.479180 [TCP Retransmission] 6112 → 64760 Len=1420
87.092606 Client ACK present; Wireshark shows tcp.analysis.ack_rtt > 1s (coincides with in-game 1500–2000 ms)
92.074249 [TCP Fast Retransmission] 6112 → 64760 Len=1420
… (many more retransmissions between ~78s and ~170s)
Display filters used (post-capture):
Latency spikes: tcp && tcp.analysis.ack_rtt > 1
Loss/congestion: tcp.analysis.retransmission or tcp.analysis.fast_retransmission or tcp.analysis.lost_segment
Capture scope: Focused on Blizzard ports (e.g., 6112). Name resolution disabled; relative seq numbers enabled.
What I need help with
Please check the health/load of the instance(s) serving my session and the peering/routing between your EU infrastructure (GCP) and my ISP.
If possible, move my session to a different instance/realm and share IP/CIDR ranges to validate from my side.
Let me know if you need the full .pcapng and exact timestamps—I can provide them along with pathping outputs.
Thank you for looking into this. From my tests, the problem isn’t my PC, NIC, or home router (all clean and sub-millisecond locally); it’s reproducible as server-side packet loss / congestion on the game flow.
Best regards