Subdomain hosted on AWS sometimes doesn't resolve towards 8.8.8.8

Recently we've been having sporadic trouble resolving our subdomain towards 8.8.8.8, so we created a second subdomain for testing, with which we are unable to replicate the issue.

Original subdomain - 42894078.sip.teltel.io - fails occasionally through 8.8.8.8 with the error: server can't find 42894078.sip.teltel.io: NXDOMAIN

Testing subdomain - 42894078.xip.teltel.io - resolves fine through 8.8.8.8

xip.teltel.io is configured exactly as sip.teltel.io.

The only difference is that xip.teltel.io is hosted by DigitalOcean, but sip.teltel.io by AWS.

Do you know why one works but the other doesn't?

All the points from the answer to your previous question still apply:

• No NS records in your sip.teltel.io domain
• The names ns1 and ns2.teltel.io still resolve to the same IP address - and it seems to be AWS IP EC2 instance address.

So it looks like you are running some DNS software (bind?) on an AWS EC2 instance - correct? I guess that's where the problem is - your bind may be misconfigured, rate-limited, DNSSEC inconsistencies or something else is at play here.

I suggest you migrate your zone sip.teltel.io to AWS Route53 managed DNS service and be done with it. It will provide all the DNS servers for you, redundancy, and it will probably cost less than running your own instance for bind.

Otherwise 1) fix the issues with your domain and then 2) dive deep into the bind logs and see what's going on there. I'd still suggest to migrate to Route53 though.

Hope that helps :)