cloudflare
🐛 (Relatively) high idle traffic consumption
Describe the bug Cloudflared tunnel consumes ~20Mb in and ~20Mb out traffic per day, even though noone is using it. All others consumers of traffic are disabled and monitored by "nethogs". This is a very remote system with metered traffic, so 20Mb/day is noticeable in the long term.
To Reproduce Start cloudflared tunnel in default configuration. Do not send any traffic through the tunnel. Monitor how much in/out traffic is consumed on the NIC
Expected behavior Very little traffic is consumed (<1Mb).
Environment and versions
- OS: Ubuntu 22.04.5 LTS
- Architecture: AMD64
- Version: 2025.2.0 (built 2025-02-05-1041 UTC)
Additional context In metrics I see lots of QUIC pings. It is very hard to fill 20Mb/day with pings, should be every second or so. If it's configurable - please suggest how to configure it to be less aggressive on pings (if pings are really to blame here).
process_start_time_seconds suggest that service uptime is ~4 days.
quic_client_received_frames{conn_index="0",frame_type="Ack"} 148690 quic_client_received_frames{conn_index="0",frame_type="Ping"} 148785 quic_client_received_frames{conn_index="1",frame_type="Ack"} 140806 quic_client_received_frames{conn_index="1",frame_type="Ping"} 140343 quic_client_received_frames{conn_index="2",frame_type="Ack"} 148814 quic_client_received_frames{conn_index="2",frame_type="Ping"} 148945 quic_client_received_frames{conn_index="3",frame_type="Ack"} 138480 quic_client_received_frames{conn_index="3",frame_type="Ping"} 137864 quic_client_sent_frames{conn_index="0",frame_type="Ping"} 149078 quic_client_sent_frames{conn_index="1",frame_type="Ping"} 140266 quic_client_sent_frames{conn_index="2",frame_type="Ping"} 149031 quic_client_sent_frames{conn_index="3",frame_type="Ping"} 137927
quic_client_sent_bytes{conn_index="0"} 2.5821164e+07 quic_client_sent_bytes{conn_index="1"} 2.0089163e+07 quic_client_sent_bytes{conn_index="2"} 2.5591699e+07 quic_client_sent_bytes{conn_index="3"} 1.7125344e+07
quic_client_receive_bytes{conn_index="0"} 1.2861758e+07 quic_client_receive_bytes{conn_index="1"} 8.211398e+06 quic_client_receive_bytes{conn_index="2"} 1.1776562e+07 quic_client_receive_bytes{conn_index="3"} 8.064028e+06
I see the same issue. Looking at my Unifi dashboard, and I see my cloudflared docker tunnel using 2-5kbps up and down at all times. All tunnel traffic goes to Nginx, and there is no traffic in Nginx access logs.
That comes to 400MB in the last 24 hours. My little old Wordpress site is absolutely not delivering this much traffic. It's all cloudflared.
I still see the issue in version 2025.2.0. At the moment I've implemented horrible (=easy to break the system) ICMP-based daemon which can stop and start cloudflared on demand, as ICMP via IPv6 is the only way I can access the system without cloudflared.
With cloudflared stopped I see <0.5Mb per day in/out.