Linode DNS update issue
I'm experiencing problems updating DNS using the API. The problem seems to be related to refreshing of the nameservers. Specifically, even after waiting 16 mins, there seems to be stale response.
Is this problem related to linode or any DNS cache?
Are all the linode nameservers guaranteed to be refreshed after 15 mins?
This problem is related to github certbot issue
5 Replies
tommydavidson
Linode Staff
Our servers are designed to update every 15 minutes, and it is typical that you will see DNS additions/changes happen within less than an hour. That said, it is not unheard of for DNS propagation to take as long as 24-48 hours while the records propagate across the internet. GoDaddy and Namecheap both have great articles which explain why this sometimes happens.
If this is an initial configuration, there isn't a whole lot that can be done to manually speed up the process, however if you configure your DNS records with a low TTL (Time To Live) value, then it will increase the frequency with which the records refresh, allowing the possibility for changes to propagate a bit sooner.
Is it possible for linode nameservers to return stale records after 15 (16) mins? The certbot client queries the nameservers directly and users are reporting getting stale records directly from linode nameservers.
Is there any delay between the propagation across the 5 linode nameservers?
wsmith
Linode Staff
Just curious, what version of the API are you using to update your DNS records?
@wsmith I'm currently seeing something similar with my setup that uses linode as a secondary DNS:
I changed my SOA about 60 minutes ago and am now requesting the SOA for my zone from ns1.linode.com. Every few requests I get a response with the old SOA.
From the response timing I'm pretty sure that at least one linode backend NS didn't update the zone. CF cached responses appear to come in after about 8 ms and backend hits seem to come in after about 90-160 ms. I get the old SOA for both "response groups". It seems the backend server with the stale cache is one which takes about 155 - 165 ms to respond (I'm talking to the CF cluster in FRA).