This might be an issue only Bell can fix but I’m hoping someone and Fly.io see’s this and has control of how they are advertising their routes.
From a Bell Canada connection in Toronto, Fly’s anycast IPs are being served from syd. That adds about 190ms to every connection, even to fly.io itself. A friend of mine on Bell is reporting the same issue. Other anycast services on the same line respond in 2–4ms (1.1.1.1 averages 3.7ms, 8.8.8.8 averages 2.4ms).
- Client: 184.146.141.200, Bell AS577, Toronto
- ping 37.16.18.81: averages 191.6ms, 0% loss
- Request IDs: 01M3YM0WCG63J1ZM75NW0J69MB-syd, 01M3YM0XMZQC04BCXGAG2A5798-syd
- Measured 2026-10-02 at 15:34 UTC
The traceroute stays on Bell to about 15ms, hands off to Zayo, then jumps to Asia-Pacific addresses (likely Singtel, then Optus):
8 64.230.79.73 12.8 ms Bell
15 64.125.28.46 68.0 ms Zayo
17 64.125.35.198 144.5 ms Zayo
18 203.208.143.186 266.0 ms
20 61.88.33.10 292.3 ms
22 59.154.50.18 188.7 ms
I might be shouting into the void here on the forum since there is no real support email but hoping someone at Fly can see this.
EDIT: Annnd just as I post this it resolves itself lol