Netgate Discussion Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search
    • Register
    • Login
    Introducing Netgate Nexus: Multi-Instance Management at Your Fingertips.

    tailscale slow speed

    Scheduled Pinned Locked Moved Tailscale
    5 Posts 3 Posters 1.3k Views 3 Watching
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • T Offline
      techpro2004
      last edited by

      So I have 2 networks a few miles apart. network a has a 1gb fios wan network b has a 2gb fios wan. I have tailscale setup to bridge the networks. On each network I have a identical dell t560 running apache2 on xubuntu. I am running a tool called librespeed in apache2. When I run a speedtest from a to b i get 462 down 234 up. when I run from b to a, I get 639 down 406 up. speedtest.net shows 881.56/689.32 on network a and 2325.77/1242.77 on b. why am I not getting full bandwidth on tailscale. I have nat-pmp enabled on both routers. I am running pfsense 26.03.1 on both routers. thanks, network a router is a dell i3020 network b is a dell tower desktop. both have x550-t2 nics.

      luckman212L 1 Reply Last reply Reply Quote 0
      • luckman212L Offline
        luckman212 LAYER 8 @techpro2004
        last edited by

        Tailscale is not designed to be a full wire-speed point-to-point (P-t-P) tunnel like traditional IPSEC. It is primarily a highly convenient and flexible mesh network, allowing one to access systems that might otherwise be inaccessible due to firewalls outside your control.

        For a high speed P-t-P tunnel, try IPSEC with AES128-GCM. The VPN Scaling page in Netgate's docs might be helpful as it suggests some optimal cipher settings when aiming for high performance.

        1 Reply Last reply Reply Quote 0
        • T Offline
          techpro2004
          last edited by

          tried ipsec also slow. I find I get the best performance with openvpn with dco. but still only getting 620/670 thanks

          luckman212L 1 Reply Last reply Reply Quote 0
          • luckman212L Offline
            luckman212 LAYER 8 @techpro2004
            last edited by luckman212

            You need to see what the real bandwidth is between these 2 networks so you have a realistic goal. Speedtest results are not reliable. Open port 5201 and run iperf3 -s on pfSense A, and then run iperf3 -P4 -w 256k -c 90.80.70.60 from pfSense B (change the IP to the wan IP of firewall A)

            Use pkg install iperf3 if you don't have it installed

            1 Reply Last reply Reply Quote 0
            • G Offline
              giuliafw70
              last edited by

              One extra thing to check is whether Tailscale is actually direct between the two sites and not falling back to DERP/relay. tailscale status or tailscale netcheck should make that obvious. If it is direct, I’d do luckman212’s iperf test both on plain WAN and through each tunnel, while watching CPU/interrupt usage on the pfSense boxes. If OpenVPN DCO, IPsec and Tailscale all stop in the same rough range, the bottleneck is probably the firewall path or NIC/driver/offload tuning rather than the VPN choice alone.

              1 Reply Last reply Reply Quote 0
              • First post
                Last post
              Copyright 2026 Rubicon Communications LLC (Netgate). All rights reserved.
              Privacy Policy · Cookie Policy