Tailscale Problems & Issues
Real user-reported problems, complaints, and limitations for Tailscale. Based on analysis of 14 signals.
Overall score: 75/100
Issues by Category
Tailscale exit node has no internet
Exit nodes often connect but pass no traffic because IP forwarding is off on the Linux node, the policy does not grant autogroup:internet, or a router (GL.iNet, OpenWrt, pfSense, TrueNAS) is missing NAT or firewall forwarding rules. Many forum threads describe the same symptom across router and NAS platforms.
Tailscale device offline after node key expired
User devices must re-authenticate when their node key expires; if nobody does, connections to and from that machine stop, and headless servers, NAS boxes and routers silently drop off. Some expired devices still look online in the admin console. Auth keys used for re-registration also expire after at most 90 days.
Tailscale MagicDNS not working or overwriting resolv.conf on Linux
On Linux, Tailscale rewrites /etc/resolv.conf when MagicDNS is on and no DNS manager is running, and setups with NetworkManager, Pi-hole or custom resolvers can lose name resolution or .ts.net lookups. Tailscale recommends systemd-resolved or --accept-dns=false as workarounds.
Is Tailscale down? Coordination server outages block logins
Third-party status trackers count roughly 70-90 Tailscale incidents per year, most involving the coordination server, admin console and API. Existing connections usually keep working, but new logins, new devices and config changes fail; most incidents last under an hour, with a few lasting several hours.
Tailscale on Windows not connecting after reboot
Windows users report the service running while the client says 'Tailscale is stopped' or 'Logged out', and tailnet addresses not resolving after boot until they reconnect. Restarting the service helps temporarily; some needed a clean reinstall.
Tailscale raised its business plan price from $6 to $8 per user
With pricing v4 (April 2026) the Starter plan was renamed Standard and its per-user price went from $6 to $8 a month, and billing moved from monthly active users to seats, including vacant ones. Existing plans keep their price for at least 12 months before migration.
Tailscale free plan is non-commercial only
The Personal plan is free for up to 6 users with unlimited user devices, but it is meant for non-commercial use only, includes 50 tagged resources ($1/month each beyond that), 3 ACL groups and basic SSH limited to 5 hosts. Businesses must pay per seat, and vacant seats are billed too.
Tailscale Android app keeps disconnecting or draining battery
Android users report the app stopping every 10-20 minutes, staying disconnected with 'Block connections without VPN' enabled, and battery use of around 10-15% when connected all day. Fixes that work for some (always-on VPN, unrestricted battery, disabling MagicDNS) do nothing for others.
Tailscale on iPhone won't connect or keeps restarting (VPN On Demand)
iOS VPN On Demand rules can stop Tailscale from connecting or immediately undo a manual reconnect, and only one VPN app can hold On Demand. A GitHub issue against version 1.92.3 on iOS 26.2 describes Tailscale repeatedly restarting until On Demand is turned off several times.
Why is Tailscale so slow? Connections stuck on DERP relay
When two devices cannot punch through NAT (blocked UDP, hard NAT, cloud NAT gateways), Tailscale relays traffic through its DERP servers, which add latency and cap throughput for fairness. Tailscale's own docs call this the most common cause of poor performance, and NAS, router and homelab forums repeatedly report big speed drops until a direct path is found.
Tailscale doesn't work with other VPNs like Mullvad
Tailscale's own FAQ says that in most cases it cannot run alongside another VPN without a workaround because of aggressive firewall rules, the one-VPN limit on iOS and Android, and conflicts in the 100.64.0.0/10 range. WireGuard-based VPNs such as Mullvad are a known conflict; the supported route is Tailscale's paid Mullvad exit-node add-on.
Tailscale SSH 'Tailworm' security flaws (October 2026)
In October 2026 a researcher published 'Tailworm', showing that an authorized Tailscale SSH session could act beyond its local user's privileges, including via SSH forwarding, plus bugs in Serve and Funnel. Tailscale fixed them in v1.104.4 (bulletins TS-2026-014 to TS-2026-017); exploitation needed an existing foothold on an allowed device.
Tailscale ACLs and subnet routes are confusing
Beyond the basic mesh, subnet routes need approval in the admin console plus IP forwarding, --accept-routes can break local LAN access, and access is governed by a JSON policy file of ACLs, grants and SSH rules. Reviewers say some tasks are possible but cumbersome and don't scale well, and G2's top con is access issues.
Tailscale requires a Google, Microsoft, GitHub or Apple login
Tailscale has no email-and-password sign-up: you log in through an identity provider (Google, Microsoft, GitHub, Apple, Okta, OneLogin or custom OIDC). Passkey accounts exist but only via an admin invite, which privacy-minded users find awkward.
Regret Patterns
Common reasons users regret choosing Tailscale.
Bought it as a privacy VPN
Users expecting NordVPN-style IP hiding discover Tailscale only connects their own devices; hiding their IP needs their own exit node or the paid Mullvad add-on.
Headless server silently dropped off after key expiry
A NAS or remote server stops being reachable months later because its node key expired and nobody was there to re-authenticate.
Remote access too slow for streaming or backups
Plex, file copies or backups crawl because the connection is relayed via DERP behind CGNAT or a strict firewall.
Work VPN or privacy VPN broke Tailscale
Turning on another VPN makes tailnet devices unreachable, and on phones only one VPN can run at a time.
Team bill rose after pricing v4
Business tailnets moved from $6 Starter to $8 Standard per seat, billed for vacant seats too, once migration applies.
Enabling accept-routes broke the local network
Devices with --accept-routes on the same LAN as the subnet router lose access to local devices by their LAN IPs.
When Tailscale Breaks Down
Specific scenarios where Tailscale fails or underperforms.
Both devices behind CGNAT or hard NAT
Hole punching fails, traffic stays on DERP relays with added latency and capped throughput; a peer relay or open UDP port is needed.
Running another VPN on the same device
Firewall rules and 100.64.0.0/10 conflicts drop tailnet traffic; phones only allow one VPN at a time.
Linux exit node or subnet router without IP forwarding
Clients connect but get no internet or cannot reach LAN devices until net.ipv4.ip_forward and IPv6 forwarding are enabled.
Linux host with NetworkManager, Pi-hole or custom DNS
Tailscale's DNS management collides with the existing resolver and MagicDNS or normal lookups break.
ISP or cloud provider using 100.64.0.0/10
Addresses or DNS servers in the CGNAT range clash with Tailscale IPs and the device loses internet or tailnet access.
Coordination server incident
Existing connections keep working but logins, new devices and admin console changes fail until the incident is resolved.
Captive portal Wi-Fi (hotels, airports)
Tailscale can't reach the internet until you disconnect, log in to the portal and reconnect.