This topic

Intermittent packet loss to specific Cloudflare IPs

patrickpy — Fri, 12 Jun 2026 16:38:14 GMT

Hello there, I've been having issues with some Cloudflare IPs, my setup is this: BT Full Fibre, own router (UniFi Dream Router 7), dual-stack (IPv4 + IPv6 via DHCPv6-PD)

Symptom: connections to some specific Cloudflare anycast IPs silently time out (packets blackholed, no RST/ICMP error), while neighbouring Cloudflare IPs work perfectly. It affects both address families, and oddly it's protocol-selective, strongly suggesting a per-flow (ECMP/LAG hash) issue rather than a routing or filtering problem. Tested at the same moment:

target ICMP TCP/443 TCP/80
2606:4700:3030::6815:4c17 DROP DROP DROP
2606:4700:3032::ac43:b965 OK OK OK
104.21.76.23 OK OK OK
172.67.185.101 OK DROP DROP <- ICMP fine, TCP dead, same instant

Where it drops: ICMPv6 traceroute to the dead destination gets well into the BT core and dies after 2a00:2000:73::53, before ever reaching Cloudflare peering:

traceroute6 -I -n 2606:4700:3030::6815:4c17
1 2a00:23c7:6bdb:2c00::1 3.794 ms (my router)
2 2a00:2302::1100:203:448 4.729 ms
5 2a00:2380:106::a1 7.877 ms
6 2a00:2380:101::5 8.049 ms
7 2a00:2000:2073:2::32 8.191 ms
8 2a00:2000:0:8040::6f 7.664 ms
9 2a00:2000:73::53 8.278 ms <- last response, then nothing
10 * * *

For comparison, the working Cloudflare destination from the same machine seconds later hands off to LINX and reaches Cloudflare normally:

traceroute6 -I -n 2606:4700:3032::ac43:b965
...
7 2a00:2000:2066::82 8.570 ms
8 2001:7f8:4::3417:1 11.487 ms (LINX, Cloudflare AS13335)
9 2400:cb00:21:2::1 11.627 ms (Cloudflare)
11 2606:4700:3032::ac43:b965 11.223 ms (destination)

Ruled out on my side: same failures from multiple source addresses and with all local VPN/filter software disabled; my router forwards the probes fine (they reach hop 9 inside BT). The same IPs are instantly reachable from external test nodes in five countries (check-host.net), so Cloudflare's side is healthy. Tunnelling my traffic through a VPN (I'm using Tailscale and Mullvad) makes everything work, consistent with only certain flow-hash buckets being affected.

I've also used AIs to help me debug this, but it didn't find any issues on my side.

Re: Intermittent packet loss to specific Cloudflare IPs

matthubbert100 — Sat, 13 Jun 2026 19:10:20 GMT

Hi. I'm having the same issue with Render services (dashboard.render.com and associated hosted sites).

Other users have performed extensive differential testing and isolated the issue entirely to the EE/BT core routing network.

The Problem

Any TCP connection over Port 443 (HTTPS) to Render's IP block (216.24.57.x) times out completely on EE Home Broadband.

Diagnostics & Cross-Carrier Testing

EE Home Broadband: 100% failure rate. Test-NetConnection on port 443 fails with a TCP timeout.
EE Mobile Hotspot: Degraded/Intermittent. It resolves via a NAT64 gateway (64:ff9b::), fails on the primary target, but occasionally succeeds on a fallback IP.
Vodafone Mobile Hotspot: 100% success rate. Instantly connects to the destination on port 443.
Local Settings: I have completely turned off EE Web Protection, Content Lock, and Parental Controls. This is not a local firewall or software block.

Traceroute Evidence (The Smoking Gun)

A tracert from my EE Home Broadband connection shows that traffic traverses the BT core network perfectly fine until it hits the Global Internet Access (GIA) edge network, where it dies immediately at hop 8 (166-49-135-83.gia.bt.net). It never leaves the BT network to be handed off to the destination:

Plaintext

Tracing route to gcp-us-west1-1.origin.onrender.com.cdn.cloudflare.net [216.24.57.7]
over a maximum of 30 hops:

  1     1 ms     1 ms     1 ms  ************
  2     4 ms     3 ms     4 ms  ************
  3     *        *        *     Request timed out.
  4     *        6 ms     *     62.172.102.140
  5     6 ms    14 ms     6 ms  peer7-et-3-0-3.telehouse.ukcore.bt.net [62.6.201.226]
  6     7 ms     7 ms     6 ms  transit1-xe-000.telehouse.ukcore.bt.net [194.72.16.177]
  7     6 ms     6 ms     7 ms  166-49-214-166.gia.bt.net [166.49.214.166]
  8     9 ms     8 ms     8 ms  166-49-135-83.gia.bt.net [166.49.135.83]
  9     *        *        *     Request timed out.
 10     *        *        *     Request timed out.
[...times out continuously up to 30 hops...]

Next Steps

Because this is happening deep within the gia.bt.net transit layer, this appears to be a broken BGP route or an issue at a major peering exchange point between BT/EE and Cloudflare/Render's upstream network.

Could a forum moderator please escalate this traceroute directly to the Tier 2/Tier 3 Network Engineering/NOC teams? Standard customer service handles cannot resolve backbone routing table issues.

I suspect this affects many other internet services hosted on that IP range that Cloudflare/Render uses.

Thank you.

Re: Intermittent packet loss to specific Cloudflare IPs

matthubbert100 — Sat, 13 Jun 2026 19:22:30 GMT

I'm having the same issue with Render services and associated hosted sites (eg dashboard.render.com)

There's clearly a major routing issue here and it's affecting EE/BT customers.

Tracing route to gcp-us-west1-1.origin.onrender.com.cdn.cloudflare.net [216.24.57.7]
over a maximum of 30 hops:

  1     1 ms     1 ms     1 ms  ************
  2     4 ms     3 ms     4 ms  ************
  3     *        *        *     Request timed out.
  4     *        6 ms     *     62.172.102.140
  5     6 ms    14 ms     6 ms  peer7-et-3-0-3.telehouse.ukcore.bt.net [62.6.201.226]
  6     7 ms     7 ms     6 ms  transit1-xe-000.telehouse.ukcore.bt.net [194.72.16.177]
  7     6 ms     6 ms     7 ms  166-49-214-166.gia.bt.net [166.49.214.166]
  8     9 ms     8 ms     8 ms  166-49-135-83.gia.bt.net [166.49.135.83]
  9     *        *        *     Request timed out.
 10     *        *        *     Request timed out.
[...times out continuously up to 30 hops...]

This needs to be escalated to engineers.

Routing failure on BT/EE backbone

matthubbert100 — Sat, 13 Jun 2026 21:27:52 GMT

Hey

I'm not a BT customer but have clients affected by what I believe to be a BT specific issue (also affecting EE). There is a routing issue causing an intermittent issue trying to connect to Render.com and other associated hosted sites (I think Cloudflare is affected too). This is causing a lot of stress for myself and my clients reliant on these services.

This traceroute demonstrates the issue. You can perform this test yourself (or using a site like globalping, and select BT) you will see very intermittent results.

How do I go about having this escalated to the appropriate engineering teams?

traceroute to dashboard.render.com (216.24.57.8), 20 hops max, 60 byte packets
1 _gateway (192.168.30.1) 0.447 ms 0.348 ms
2 172.16.11.176 (172.16.11.176) 2.025 ms 1.985 ms
3 * *
4 62.172.102.224 (62.172.102.224) 2.330 ms 2.289 ms
5 62.6.200.7 (62.6.200.7) 2.397 ms 2.374 ms
6 transit1-xe-0-0-1.telehouse.ukcore.bt.net (194.72.16.179) 2.342 ms 2.316 ms
7 166-49-214-166.gia.bt.net (166.49.214.166) 2.361 ms 2.372 ms
8 166-49-135-83.gia.bt.net (166.49.135.83) 2.943 ms 3.039 ms
9 * *
10 * *
11 * *
12 * *
13 * *
14 * *
15 * *
16 * *
17 * *
18 * *
19 * *
20 * *

Re: Intermittent packet loss to specific Cloudflare IPs

Edinburgh-wg — Sun, 14 Jun 2026 04:54:58 GMT

It must be your setup with UniFi Dream Router or your computer because all is working fine here.

Anyway.

Change the way ipv4 and ipv6 are negotiated, so not DHCPv6.

Also, what upstream DNS servers do you use and what happens if you use a different one?

What happens if you switch off your firewall (router/PC)?

What happens if you remove a RPi, if you have one and if you're routing traffic through it?

What happens if you use a different machine (mobile/laptop) to access sites?

You could try resetting your router to factory defaults and start with a basic setup and removing any devices in-between like a RPi etc., then gradually introducing whatever customisation you require.

Without knowing much more, it's difficult to tell, but if your VPN is working fine, then I bet on your Dream Router setup problem. You could swap your router with BT's one - that was sent to you for free - to confirm that.

I'm surprised you're trying to use AI to troubleshoot.

You'll have a better outcome by contacting professionals at UniFi to troubleshoot your UniFi Dream Router setup.

I also believe a simple setup is the best.