2

I have two AWS EC2 instance in Sydney, serving sites sitting behind CloudFlare. I'm not using ELB. Over the last week I have started seeing CloudFlare serving 522 Timeout errors fairly regularly.

CloudFlare support said:

Currently there are no signs of congestion or issues that should be causing TCP connect timeouts from CloudFlare's network. After running tests from SYD/MEL, I am seeing that certain ranges are experiencing timeouts when attempting to connect to the site's origin:

SYD: 108.162.250.106: nc: connect to (MY IP) port 80 (tcp) timed out: Operation now in progress
MEL: 162.158.2.147: nc: connect to (MY IP) port 80 (tcp) timed out: Operation now in progress
MEL: 162.158.2.204: nc: connect to (MY IP) port 80 (tcp) timed out: Operation now in progress
MEL: 162.158.3.42: nc: connect to (MY IP) port 80 (tcp) timed out: Operation now in progress

I recommend confirm with your host that CloudFlare IP ranges are whitelisted and not being rate-limited from within their network.

I have completely open http and https limits in the Security Group. I'm not aware of AWS blocking or rate limit by IP without using ELB, is this the case? If they do, how can I go about checking/updating the list of CloudFlare IPs known to AWS? I didn't set up any CloudFlare IPs originally.

One server is running Ubuntu 14.04 with ServerPilot, the other is running the AWS AMI with LAMP stack (standard AFAIK, I didn't set that one up). I have asked ServerPilot whether their NGINX or iptables set-up could be causing this, but since it's also happening with a sever not managed by ServerPilot I doubt that's the cause. Both instances are in different Availability Zones.

Thanks

Josh B
  • 21
  • 3

0 Answers0