Spotty connection for multiple users in another country

Hello i have a demo app running that has a spotty connection to Canada testers. The demo has been up for a week and today is the first day multiple people started noticing slow site load times. Everything is fast (B 83 performance via pingdom) in the US. How do i troubleshoot that when it’s on a subdomain and I’m not in Canada? The app runs great in the U, London, and Tokyo but is spotty in Canada. I’m using the free version of cloudflare.

Also in Canada and I’m seeing packet loops in trying to get requests to my workers. Any chance you could get a tcptraceroute to see if it’s the same?

Also seeing issues from Canada, both on my home ISP (Rogers) and from Linode’s datacenter in Toronto.

[email protected]:~# wget https://pages.cloudflare.com/
--2021-03-27 20:23:55--  https://pages.cloudflare.com/
Resolving pages.cloudflare.com (pages.cloudflare.com)... 2606:4700:134:6cfa:7196:3bd9:93e8:9233, 172.64.81.132
Connecting to pages.cloudflare.com (pages.cloudflare.com)|2606:4700:134:6cfa:7196:3bd9:93e8:9233|:443... failed: No route to host.
Connecting to pages.cloudflare.com (pages.cloudflare.com)|172.64.81.132|:443... failed: No route to host.

Issues from Bell to Toronto Cloudflare.

The issue seems to be resolved. I am interested in what was causing the issue in the first place.

I can’t get to the 172.64.80.0 network either. Google’s DNS (8.8.8.8) gives me other addresses I can connect to. You might want to give it a show. I tweeted my findings to their help acct, but haven’t heard back.

$ nslookup pages.cloudflare.com 1.1.1.1
Server:		1.1.1.1
Address:	1.1.1.1#53

Non-authoritative answer:
Name:	pages.cloudflare.com
Address: 172.64.81.40

$ nslookup pages.cloudflare.com 8.8.8.8
Server:		8.8.8.8
Address:	8.8.8.8#53

Non-authoritative answer:
Name:	pages.cloudflare.com
Address: 172.67.144.27
Name:	pages.cloudflare.com
Address: 104.21.87.149
1 Like

How do i resolve? Unless cloudflare reports it as a issue, I don’t have any facts to relay about the cause of the problem or the origin.

I am also seeing better results, but still not 100%. Issue for me seems less frequent but still present.

The issue doesn’t appear to reside on the cloudflare side, looks to be a route loop in twelve99 which looks to be a major backbone provider for Toronto. You can test routes here:
https://lg.twelve99.net/

Not an networking expert, but does the appearance of the torontointernetexchange.net address on both hop 5 and 6 indicate a loop?

traceroute to pages.cloudflare.com (172.64.81.82), 30 hops max, 60 byte packets
 1  gw-li1215.linode.com (45.79.116.1)  0.586 ms  0.535 ms  0.521 ms
 2  10.215.64.6 (10.215.64.6)  0.721 ms  0.713 ms 10.215.64.5 (10.215.64.5)  0.697 ms
 3  10.215.32.36 (10.215.32.36)  2.064 ms 10.215.32.32 (10.215.32.32)  2.058 ms  2.051 ms
 4  10.215.32.49 (10.215.32.49)  0.698 ms  0.704 ms  2.014 ms
 5  cloudflare.ip4.torontointernetxchange.net (206.108.34.208)  1.997 ms  3.153 ms  3.145 ms
 6  172.64.81.82 (172.64.81.82)  0.558 ms cloudflare.ip4.torontointernetxchange.net (206.108.34.208)  1.767 ms 172.64.81.82 (172.64.81.82)  0.856 ms

tcptraceroute always seems to run into issues (!N) at the same IP: 172.70.45.1:

Running:
        traceroute -T -O info pages.cloudflare.com
traceroute to pages.cloudflare.com (172.64.81.224), 30 hops max, 60 byte packets
 1  gw-li1215.linode.com (45.79.116.1)  0.172 ms  0.152 ms  0.141 ms
 2  10.215.64.6 (10.215.64.6)  0.220 ms 10.215.64.5 (10.215.64.5)  0.227 ms  0.221 ms
 3  10.215.32.32 (10.215.32.32)  2.146 ms  2.140 ms 10.215.32.36 (10.215.32.36)  0.770 ms
 4  10.215.32.49 (10.215.32.49)  0.651 ms  1.858 ms 10.215.32.51 (10.215.32.51)  0.385 ms
 5  cloudflare.ip4.torontointernetxchange.net (206.108.34.208)  1.052 ms 172.105.31.9 (172.105.31.9)  0.355 ms  1.467 ms
 6  * * cloudflare.ip4.torontointernetxchange.net (206.108.34.208)  2.189 ms
 7  108.162.218.225 (108.162.218.225)  15.525 ms 172.70.45.34 (172.70.45.34)  14.516 ms 172.70.45.52 (172.70.45.52)  14.637 ms
 8  108.162.218.171 (108.162.218.171)  15.223 ms 108.162.218.37 (108.162.218.37)  15.529 ms 108.162.218.1 (108.162.218.1)  16.681 ms
 9  172.70.45.1 (172.70.45.1)  15.765 ms !N  15.716 ms !N  15.459 ms !N

Just as I was reading this thread, this popped up in my feed:
https://community.cloudflare.com/t/network-performance-issues-in-toronto-on-and-montreal-qc-canada/255934

2 Likes

I noticed the following in their post “for Free Traffic.”

Is cloudflare performance quality / network routing different depending on what spend level you are at? The reason I ask is I manage sites that use both free and the first pay tier. The free tier sites have the most reported problems, almost to the point that sometimes I feel like I’m reducing my quality of service (uptime + page speed) in exchange for less server load (cloudflare resource caching, DDoS protection, custom down pages). In theory the trade off might only be worth it on sites with a lot of traffic. But if the performance and uptime isn’t the same on all tier’s, It’s hard to tell what the best fit is beyond the brand.

Over the last year, all network problems reported by my clients were the result of cloudflare problems. 1 was on the entry level paid tier and the rest were on free tiers. All confirmed by cloudflare after investigation. Do any of you have free and paid tiers where you also monitor performance using third party or self hosted tools? Do you also have quality differences in the tiers?

I created a zabbix server (looking for alternatives) to help me monitor page speed since I think most site responsiveness problems result in users leaving the site rather than being reported. I don’t have enough data yet to know if the quality is different or if I should change the way I use cloudflare,

This topic was automatically closed 5 days after the last reply. New replies are no longer allowed.