From: Luis S. <lui...@gm...> - 2008-09-01 11:24:20
|
Hi all! When sending NXDOMAIN and NXRRSET from a nameserver to a client, the response, in the authoritive section, might contain the SOA record of the authoritive zone. Using the dnsjava Cache, when looking for this records, the SetResponse will not contain this Record. Is this feature planned to be implemented or you just search again for the SOA record if you want to send it in the response? Kind Regards, Luis |
From: Luis S. <lui...@gm...> - 2008-09-01 17:17:00
|
Hi! Sorry for this last email since I noticed that in this cases you case the negative element with the soa record. But I also noticed that you used the ttl value from the soa record instead of the minimum. According with the rfc 2308 it says that the minimum value is used in these situations: "The remaining of the current meanings, of being the TTL to be used for negative responses, is the new defined meaning of the SOA minimum field." you see the opposite in the addNegative method in Cache.java: addNegative(Name name, *int* type, SOARecord soa, *int* cred) { *long* ttl = 0; *if* (soa != *null*) ttl = soa.getTTL(); ... } Do you agree? On Mon, Sep 1, 2008 at 12:24 PM, Luis Silva <lui...@gm...> wrote: > Hi all! > When sending NXDOMAIN and NXRRSET from a nameserver to a client, the > response, in the authoritive section, might contain the SOA record of the > authoritive zone. Using the dnsjava Cache, when looking for this records, > the SetResponse will not contain this Record. Is this feature planned to be > implemented or you just search again for the SOA record if you want to send > it in the response? > > Kind Regards, > Luis > |
From: Brian W. <bwe...@xb...> - 2008-09-01 17:23:57
|
On Mon, 1 Sep 2008, Luis Silva wrote: > Hi! > Sorry for this last email since I noticed that in this cases you case the > negative element with the soa record. But I also noticed that you used the > ttl value from the soa record instead of the minimum. According with the rfc > 2308 it says that the minimum value is used in these situations: > > "The remaining of the current meanings, of being the TTL to be used > for negative responses, is the new defined meaning of the SOA minimum > field." > > you see the opposite in the addNegative method in Cache.java: > > addNegative(Name name, *int* type, SOARecord soa, *int* cred) { > *long* ttl = 0; > *if* (soa != *null*) > ttl = soa.getTTL(); > ... > } > Do you agree? I don't see a problem here. The cache is caching the records as transmitted, which looks fine. If a caching server is planning to use the records from the cache, it might need to do additional work, just as any other caching server would do. Brian |
From: Luis S. <lui...@gm...> - 2008-09-01 17:30:20
|
You are responding to my first or second email? Regarding the first I think that it might help insert the negative element inside the SetResponse. Regarding the second, the ttl value must be the minimum value and not the ttl value from the soa record. Kind Regards, Luis On Mon, Sep 1, 2008 at 6:23 PM, Brian Wellington <bwe...@xb...> wrote: > On Mon, 1 Sep 2008, Luis Silva wrote: > > Hi! >> Sorry for this last email since I noticed that in this cases you case the >> negative element with the soa record. But I also noticed that you used the >> ttl value from the soa record instead of the minimum. According with the >> rfc >> 2308 it says that the minimum value is used in these situations: >> >> "The remaining of the current meanings, of being the TTL to be used >> for negative responses, is the new defined meaning of the SOA minimum >> field." >> >> you see the opposite in the addNegative method in Cache.java: >> >> addNegative(Name name, *int* type, SOARecord soa, *int* cred) { >> *long* ttl = 0; >> *if* (soa != *null*) >> ttl = soa.getTTL(); >> ... >> } >> Do you agree? >> > > I don't see a problem here. The cache is caching the records as > transmitted, which looks fine. If a caching server is planning to use the > records from the cache, it might need to do additional work, just as any > other caching server would do. > > Brian > |
From: Brian W. <bwe...@xb...> - 2008-09-01 17:36:50
|
On Mon, 1 Sep 2008, Luis Silva wrote: > You are responding to my first or second email? I'm responding to the text that I quoted. > Regarding the first I think that it might help insert the negative element > inside the SetResponse. Regarding the second, the ttl value must be the > minimum value and not the ttl value from the soa record. Adding the SOA record to the NXDOMAIN/NXRRSET SetResponse sounds reasonable, but I believe that any server code using the SetResponse can deal with TTL issues itself. If you want to send a patch to implement the first part, you can. Otherwise I might implement it at some point, but I don't expect to have any time to work on dnsjava for a few weeks. Brian |