Let's say I have a server (DNS and other), myserver.com. Now I register a domain, mydomain.com, and set it's NS at the registrar to myserver.com - it is therefore the authoritative server, if there is any such thing.
In the authoritative records for mydomain.com, can I set the NS to ns.mydomain.com?
I have two domains set up like that, one works, the other one seems reluctant to propagate. So I'm wondering if there is something wrong with that - I mean how can you resolve the name of the NS when you need to resolve the name of the NS to resolve the name of the NS...
And, If yes, how come parallels plesk sets them automatically in this way?
Ps: there is an A record for ns.mydomain.com on that same server, pointing to the proper IP
There's a solution for this problem - it's called "glue records", i.e. A records hosted in the parent zone that contain the IP addresses of the name servers.
See http://en.wikipedia.org/wiki/Domain_Name_System#Circular_dependencies_and_glue_records
Why would you want to set the NS record for the "mydomain.com":
to "myserver.com" in the delegation record that goes into the parent zone (com.), but
to "ns.mydomain.com" at the zone apex (inside the mydomain.com. zone)
? This creates an inconsistency (two different DNS servers answer the same question with two different answers) without any apparent benefit. You should try to help the DNS system as a whole issue consistent answers.
Unless you have a good reason to make the DNS inconsistent, you should decide what the correct, canonical name for your nameserver is, and publish that name in the NS record both in the delegation and at the zone apex for "mydomain.com".
That being said, it will still work:
If a recursive resolver which does not yet know anything about "mydomain.com" asks about it, it will be told by the gTLD servers to go look at "myserver.com". The gTLD will also issue A and AAAA glue records to help find "myserver.com", but even if they don't, you have A and AAAA records for "myserver.com" in the "myserver.com" zone file (right?).
If a recursive resolver which wants to refresh its cache for the "mydomain.com" NS record, it may query the authoritative server it already knows about. This server will answer that the nameserver is "ns.mydomain.com", with a glue record. This is different from what it had in its cache before, but ultimately it will map to a server with the same IP address.
As for "parallels plesk", I know nothing about that.
Related
I am looking at the DNS setup for a company, call them bar.com. They have a zone file for a subdomain, so foo.bar.com has a few bind servers; so far so good. The thing that strikes me as strange is that the NS record for foo.bar.com points at ns.foo.bar.com. That seems weird to me - the authoritative name servers for a zone are contained within the zone itself. That feels like a snake eating its own tail, the sort of thing that could end badly. Is that standard practice or is it indeed weird? The setup, for clarity:
NS record for foo.bar.com -> ns.foo.bar.com
A record for ns.foo.bar.com -> IP of ns1.foo.bar.com
A record for ns.foo.bar.com -> IP of ns2.foo.bar.com
A record for ns.foo.bar.com -> IP of ns3.foo.bar.com
This is a standard practice. Many large online businesses have their own nameservers delegated in this way. When registering a domain you can choose to set the nameservers to hostnames from another domain (a public DNS provider) or you can set them to names inside your domain. When registering with nameservers as part of the domain you must also specify the IP address of the nameservers as glue records (A or AAAA records created in the parent zone). These records are returned in the ADDITIONAL section of the dns response from the parent zone. This allows a resolver to find your nameservers and then query your records.
I'm trying to learn as much as possible about DNS, and so far I've read most of:
http://www.zytrax.com/books/dns/ch8/soa.html
and all of:
http://computer.howstuffworks.com/dns.htm
I understand that SOA and NS records contain info about the authoritative name server for a domain, but as these are just DNS records, how does the rest of the world even know where to get them?
I assume it starts at the top-level-domain (.COM .NET .ORG, etc) servers. So they must contain a SOA record for my domain? If so, how does that get there? I imagine only registrars like GoDaddy and Network Solutions are able to update those? If they contain a SOA record, why does my DNS server (that I host), need one also? I think there must be something, maybe in the domain registration records (outside of DNS?), that I'm missing.
I think I've got a pretty good understanding of most parts of the DNS system, after reading lots of articles.. but I haven't found any that answer this part, in a way that I understand it.
For example, GoDaddy and Network Solutions both let me change different options (in their web UI) to "host my own DNS server". If these options remove them from the process, so DNS servers never need to query them again, and instead query my server directly (this is what I want, no dependency on GoDaddy/NS)... when I make these changes, what (at the DNS level or otherwise) is GoDaddy/NS doing? Are they asking the top-level-domain servers to update some DNS records for my domain?
Short answer is yes. Godaddy will take care of updating your the TLDs for you. Your other assumptions are also correct except for one small detail...
DNS starts with the "root" domain and then goes to the TLDs (top level domains).
www.somedomain.com. actually brakes down like this:
. The root name servers
com - The TLD name servers
somedomain - Your NS servers
www - The host portion of the dns name.
Setting up your own DNS server is a great way to understand DNS better. Good luck!
So they must contain a SOA record for my domain?
Multiple answers possible, depending on the scenario:
delegated 2nd level:
No; the start of authority of your domain is usually at your level, so the SOA for you.cf is (only) in your nameservers. Same as the SOA for .cf is only in the nameservers of .cf, and not in the root-servers. In case your nameservers are within the same domain (i.e. ns1.you.cf in case of you.cf) then glue records are needed. This means that registry that's operating the TLD's nameservers will insert a A and/or AAAA record with the IP of ns1.you.cf in the TLD zone. Normally you (the registrant) set this in the interface of your registrar.
undelegated 3th level:
No; if your domain was sub.you.cf - but sub.you.cf wasn't delegated (no NS records exist for sub.you.cf) then the SOA is probably at you.cf. Unless...
undelegated 2nd level:
Yes; if you register a domain without having it delegated (no NS records exist for you.cf) then the SOA is at the .cf nameservers. Although most registries run delegation-only zones, some don't. Example: when you use url-fwd'ing in .cf. The SOA then is at the cf level.
Is DNS lookup time slowed down by the use of subdomains and CNAME? As I understand it, if a client wants to lookup for example rweb.stat.ucla.edu at least four steps are needed:
[registrar] NS ucla.edu ==> ns2.dns.ucla.edu
[ns2.dns.ucla.edu] NS stat.ucla.edu ==> dns.stat.ucla.edu
[dns.stat.ucla.edu] CNAME rweb.stat.ucla.edu ==> id-86-243.stat.ucla.edu
[dns.stat.ucla.edu] A id-86-243.stat.ucla.edu ==> 128.97.86.243
Does this mean the client has to connect to each of these servers when visiting the page? Or do the DNS servers connect to each other and directly resolve (and cache) rweb.stat.ucla.edu ==> 128.97.86.243? Also it is not clear to me how much work is involved in finding the initial record for the root domain?
First of all, the registrar is never involved in the actual DNS lookup. Registrars (and indeed, registries) are administrative entities outside the scope of the DNS protocol. The actual lookup process looks more like this:
Ask root servers for A of rweb.stat.ucla.edu..
They won't tell you, but they'll point you to servers for edu..
Ask the edu. servers for A of rweb.stat.ucla.edu..
They won't tell you, but they'll point you to servers for ucla.edu.
Ask the ucla.edu servers for A of rweb.stat.ucla.edu..
stat.ucla.edu could be a zone that is delegated to yet another set of nameservers, in which case you'll get a referral answer for the third time.
But let't say it's not.
You get an answer: rweb.stat.ucla.edu. has CNAME id-86-243.stat.ucla.edu.
Now you have to start all over again and look up id-86-243.stat.ucla.edu...
...except that because the target of the CNAME is inside the same zone as the original record (or, in general, inside any zone that happens to be served from the same nameserver), the nameserver will helpfully give you the A record for id-86-243.stat.ucla.edu in the additional section of the DNS response. The resolver will notice this and won't have to run after the A record itself.
Because resolvers cache answers, many queries can be answered from the cache and don't actually have to be asked every time. Especially the first few steps of the query are all likely to be already cached by the resolver, except immediately after it starts up with a cold cache. So the resolver can probably usually start at step 3 (because it remembers what the nameservers for ucla.edu. are from its cache) or at least at step 2 (because it remembers what the nameservers for edu. are) if it doesn't already know the final answer.
So the short answer to your question is yes, using CNAME makes more work for the resolver and makes the resolution process take longer. But in practice it doesn't matter very much because the resolver's cache is so helpful. And if the CNAME's target is in the same zone as the CNAME record itself, it matters even less because of the helpful additional record.
I have inherited a web server that is hosting 5 websites for my client. Call them domian1, domain2, etc I just discovered that all the domain nameservers for all 5 domains are set to ns1.domain1.com and ns2.domain1.com. The single server is running the DNS for all the domains including domain1.com. ns1 and ns2 are both pointing to the same web server.
Aside from the fact that there is no redundancy, and the domain1 name servers are using the DNS to resolve their own IP's, why would anyone do this? Am I missing something?
There are two options when creating NS records for zones:
1) Set the NS record of each zone to point only to itself. Hence, domain1.com would get ns1.domain1.com, etc. The advantage of this is that the remote site doesn't need to do a cross reference to somewhere else and go look it up too. EG, if you have domain1.com's NS records pointing to ns1.domain2.com, then a lookup of the NS records for domain2.com have to be checked too to ensure it has the right location to go lookup where ns1.domain2.com really is. You could imagine the case where domain2.com's NS records point to domain3.com's name servers... This is obviously inefficient and results in a lot of unneeded chasing. So... pointing entirely internal seems like a no-duh, right! Less chasing! But... it also means you need to keep com's notion of your name servers in sync with your notion of your name servers, and when you add or remove them and/or change the IP addresses, you need to notify your com (through your registrar) that things have changed. (tech speak: update com's notion of your glue records).
2) Add an NS record pointing to an external server. This is common for server farms that sell you DNS services as part of their transaction as your registrar (ie, where you went to go buy domain1.com). They set your NS record to something like "ns1.godaddy.com". In your case, the previous zone owner set the NS records to all point to the domain1.com zone. This is actually helpful when you expect to change your address in the future. Rather than have to go change the IP address in all 5 of your zones, you only change it in domain1.com's ns1.domain1.com record and you're good to go. The other zones don't need to be touched. Yay! It's even more yay-full when you are managing 100 zones.
So, there isn't a right or a wrong here... It's a trade-off and different administrators do different things. Feel free to change it to the other model if you don't mind the zone-editing maintenance if you ever change anything. Personally, it's what I do when possible too: I like them internally self-contained. But then, that's also when most people fail to update the parent's glue records to match and there are tons and tons of zones in the world that are out of sync for exactly this reason: "oh, I'll do that tomorrow".
I setup a couple of nameservers and updated my domain to use them, and as far as I can tell everything went fine and the nameservers have been updated, or so says every whois and dnstools type site ive used, (intodns, who.is etc are all saying the same thing: the new nameserrvers are in effect, and the site points to the new ip just fine). Problem is that The site is not showing up, and dig tells me that the old ip/nameservers are still effective.
In my DNS Records I have:
domain. A IN NS ns1.newnameserver
domain. A IN NS ns2.newnameserver
ns1 IN A newipaddress
ns2 IN A newipaddress
domain. IN A newipaddress
I'm very short on time and haven't found anything on the interweb, so any help would be much appreciated
The old IP address is probably being cached by the server you queried. First of all, check that BOTH your new authoritative nameservers are publishing the correct address by querying them directly with dig:
dig #ns1.newnameserver domain. a
dig #ns2.newnameserver domain. a
Assuming those queries give correct answers, dig some other servers that aren't:
dig domain. a # Use the system's default resolvers
dig #8.8.8.8 domain. a # Use Google's public resolver
dig #some.other.ip.address domain. a
If it gives the old answer, look at the TTL. That's the numeric field listed in the answer just after the name and before "IN". That's how many seconds you have to wait until the server you queried discards its cached data and will query the authoritative servers again.
Ask those same nameservers where they think "domain." is delegated:
dig domain. ns # Use the system's default resolvers
dig #8.8.8.8 domain. ns # Use Google's public resolver
dig #some.other.ip.address domain. ns
You want to see 2 NS reocrds for "domain.", one pointing to "ns1.newnamserver" and the other one to "ns2.newnameserver", but the resolvers likewise cache that information so they might still have the old nameservers. If so, look at the TTL on those NS records too. If the TTL on those records is longer than the TTL on the A records, those resolvers may still go to the old nameservers to get "domain."'s A records even when their currently cached copy expires... so you may need to wait for that TTL to expire first, and then for the TTL on the actual A record to expire again!
Another thing you can do is query some of the authoritative nameservers for the PARENT domain of your domain to see if they are indeed delegating it to "ns1.newnameserver" and "ns2.newnameserver". This will verify that the delegation in DNS matches what's in WHOIS.
dig com. ns # If your domain's parent domain is "com."
dig #<one-of-the-servers-that-resulted-from-that-query> domain. ns
Again, you want to see 2 NS reocrds for "domain.", one pointing to "ns1.newnamserver" and the other one to "ns2.newnameserver".
If the old nameservers are still running, either:
make sure they aren't, or
make sure they've also got the new zone data
Some people will still be talking to the old nameservers, and until they either stop answering or give the right answer, they won't learn the new nameservers from the parent zone.