Score:0

How will requests to DNS servers be processed?

je flag

I plan to move the external ns zone outside the country, 3 entries at the domain registrar will indicate external addresses(outside the country) and 1 entries will indicate my local ns(inside the country). When querying the registrar, he gives out the results randomly, that is 4 records are swapped.

Questions:

  1. Let's say one NS server will always be in the disabled state, in this case, if a request to domain registrer returns a response, where the disabled NS is the first in the list, what will be the result of the client?.

  2. I assume that the next NS will be requested, and so on in turn. If it true, can you please tell me the RFC number where this behavior is documented (I don’t know how to correctly make a request in google).

Patrick Mevzek avatar
cn flag
Terminology wise, you are mixing "domain name registrar" and "DNS provider". One entity can do both, but the jobs are completely different. Your question is solely related to DNS management, and not domain name registrations.
Patrick Mevzek avatar
cn flag
"Let's say one NS server will always be in the disabled state" That is not how DNS is supposed to work. You have multiple `NS` records to handle redundancy and fail over but it is not good practice to publish a record that you know will never work, because the server is unreachable. If other nameservers are working then your DNS will work, albeit with delays. Yet your DNS setup is still fragile and giving a false impression of fail over by looking just at `NS` records.
Tommiie avatar
bw flag
Why will one NS always be disabled (define disabled). Perhaps you want to make it a hidden master instead?
Score:0
cn flag
Bob

Wikipedia has quite a useful list with all RFC's https://en.wikipedia.org/wiki/Domain_Name_System#RFC_documents

As an intro into the original intended behavior of the DNS spec this is probably the section you're after: https://datatracker.ietf.org/doc/html/rfc1034#section-5.3.3

As for what will be the result for queries sent to a disabled name server:

I would expect a difference in behaviour when the name server is powered off (resulting in a connection time-out) , compared to the server being powered on, running a DNS server but not responding for your domain i.e the server returns a valid DNS response with a DNS error code such as REFUSED, SERVFAIL or NXDOMAIN .

The latter error will be returned to the client, but the first scenario should result in the query being asked by the resolver to (one of) the other authoritative name servers for your domain.

As Patrick commented : what will then actually happen (in edge cases and failure scenarios) depends on the specific software and version.

Patrick Mevzek avatar
cn flag
How resolvers choose which nameserver to query is complicated and not related to IETF RFCs. As such it depends on the software used and the version which basically means the final outcome is undefined for all cases. In summary and in theory, a resolver will try all nameservers at beginning before gathering enough data on how they reply and how fast so that at the end (ongoing resolutions) the fastest nameserver is queried almost always but not exactly always (to be still able to switch over another one if one becomes suddenly faster).
Patrick Mevzek avatar
cn flag
Wikipedia list of RFCs may be good but also hard to read and failing to be exhaustive. This link to unbound documentation gives exactly which RFCs this recursive resolver is implementing, giving a more precise on the field idea of what happens: https://unbound.docs.nlnetlabs.nl/en/latest/reference/rfc-compliance.html
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.