Dynamic DNS fails

[Volker Wysk]

Am Montag, den 22.03.2021, 08:21 -0400 schrieb Ken D'Ambrosio:
> On 2021-03-22 07:42, Volker Wysk wrote:
> > Am Montag, den 22.03.2021, 08:15 +0100 schrieb Bo Berglund:
> > > Does your ddns service update the DNS servers around the world?
> 
> Respectfully, that's not a thing, and not the way DNS works.  DNS 
> doesn't update *anything* with its hostname entries (except its own 
> domain's servers).  It certainly doesn't "update the DNS servers around 
> the world."  What it *does* do is accept queries.  The way things 
> usually work:
> 
> * Client host makes a request of (say) bar.foo.com to its DNS server 
> (often the home router)
> 
> * The home router forwards the request to *its* DNS server, e.g., 
> Google's 8.8.8.8, your ISP's DNS server, etc.  This is usually what's 
> termed a "caching server," for reasons that will be obvious in a moment.
> 
> * 8.8.8.8 or the ISP's server, a caching server, maintains caches of 
> entries so that it doesn't have to do a full query each time.  (Windows 
> clients also do this by default; Linux doesn't, but can with "nscd" -- 
> Name Services Caching Daemon -- installed.)  If it has an unexpired 
> cache entry for the host query, it responds with that, and you're done.  
> There's one caveat, here, though -- which is why I'm bothering to type 
> all this up: the failed lookup.  If, for some reason, a caching server 
> attempts a lookup and fails, it'll cache *that*, too, and respond with a 
> failed lookup until the cache entry hits its expiration time, which can 
> vary from caching server to caching server.  I suspect that this is what 
> happened.
> 
> * What if the caching server *doesn't* have an entry?  It then does a 
> query against the DNS root servers to find out where the DNS server for 
> the domain "foo.com" resides, and then queries the foo.com's server 
> directly for host "bar", the server responds with the appropriate DNS 
> entry, which is then cached by the caching server, and pushed down the 
> chain to the client.
> 
> So, my guess: Volker's intermediate caching server tried to do a resolve 
> against his dynamic DNS provider's server, and it failed for some reason 
> -- perhaps maintenance, a routing glitch, whatever.  That failure then 
> got cached by the caching server, and was there, repeatedly failing for 
> him, until the entry expired.

Yes, this matches the behavior I've observed. When doing a "ping vwysk.now-
dns.net", first it takes a long time to respond (some five or ten seconds).
After that, doing it again, there is no noticeable delay. It stays like this
for another five or ten seconds, then it's back to the long delay again.

By the way, here is a good DNS dictionary:
https://www.plixer.com/blog/overview-of-dns-protocol-part-1-of-3/

Regards,
Volker

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <https://lists.ubuntu.com/archives/ubuntu-users/attachments/20210323/bb068da3/attachment.sig>