2

I have a site which is responding with DNS_PROBE_FINISHED_NXDOMAIN at random places around the globe which has been getting worse for the last 16 hours.

I have a Google Cloud CDN attached. everything was working fine for months, until yesterday this DNS_PROBE_FINISHED_NXDOMAIN response started getting sent and the site is randomly unreachable. I have no idea why, any thoughts on what to look into?

Setup: Godaddy registrar => Google Cloud Network Load balancer IP => Cloud Run Static node Server (with CDN Turned on)

The Site is https://smodin.io , and you can see random nodes down across the globe (https://www.whatsmydns.net/#A/smodin.io).

I have determined that the site is unreachable on xfiniti, but is reachable from the phone hotspot.

Kevin Danikowski
  • 4,620
  • 6
  • 41
  • 75
  • 1
    Your site is available via **253.39.117.34.bc.googleusercontent.com** using curl. This means that you have a configuration problem at the load balancer. – John Hanley Sep 02 '21 at 15:10
  • Thank you John, I will look into this. I didn't do any load balancer updates which is why I'm so perplexed. Thank you! – Kevin Danikowski Sep 02 '21 at 15:18
  • 2
    This could be a transient problem that will self-correct. Try creating a new HTTP and HTTPS frontend using a subdomain to see if that works (e.g. www2.smodin.io). Do not use a CNAME for this test, use an A record. – John Hanley Sep 02 '21 at 15:37
  • @JohnHanley based on your advice, using a test domain of mine on the same project with the same issue (app-translation.com), after adding the www2 with load balancer I was able to successfully get it work on more nodes. I'm going to try now with the main domain. – Kevin Danikowski Sep 02 '21 at 16:05

1 Answers1

0

Thank you to @JohnHanley for the suggested insight.

Looking at https://www.whatsmydns.net/#A/smodin.io it went from having about 10-15 nodes out randomly across the globe, to 2.

Here is what I did, there might be steps that are unnecessary though.

  1. Add A record to godaddy to point www2 to the current IP your load balancer is listening at (this step is necessary I found, I didn't try pointing to the new static IP I created though).
  2. Create a new load balancer for https. This includes creating an SSL cert. I also didn't use an ephemeral IP, I created a brand new static IP (might have been unnecessary)
  3. point it to the client of choice (I didn't duplicate my entire settings, I just put my main client as the receiver for all requests, it's quite possible you can put an empty bucket instead)
  4. Create it.... wait about 5-10 minutes.... you should start seeing your nodes come back up.
  5. after about an hour has passed and DNS has propagated, you can delete the A record and associated resources.
Kevin Danikowski
  • 4,620
  • 6
  • 41
  • 75