3 ms·
It'd be nice to have a "backup TTL" included, to allow sites to specify whether and how long they wanted such caching behavior. Also, that cache would need to
by JoshTriplett 10y ago
It'd be nice to have a "backup TTL" included, to allow sites to specify whether and how long they wanted such caching behavior.
Also, that cache would need to only kick in when the server was unreachable or produced SERVFAIL, not when it returned a negative result. Negative results returned by the authoritative server are correct, and should not result in the recursive resolver returning anything other than a negative result.
- bhauer 10y ago> Also, that cache would need to only kick in when the server was unreachable or produced SERVFAIL, not when it returned a negative result. Negative results returned by the authoritative server are correct, and should not result in the recursive resolver returning anything other than a negative result. Precisely. I am not suggesting any change to how a caching resolver comprehends valid responses from the authoritative server for a domain. For example, if the authoritative server says, "No such domain," then the domain is understood to be gone. At that point, the domain being gone is in fact the last-known-good resolution.
- spc476 10y agoIt might be a stretch to use the information, but the SOA RR does contain an EXPIRE field, defined as "A 32 bit time value that specifies the upper limit on the time interval that can elapse before the zone is no longer authoritative." It's an additional request, but the SOA RR does contain the type of information you are asking for.
- Godel_unicode 10y agoThis is why TTL is a variable. If you want to have your records last longer, you can. If you want to later shorten them, you can. People screw DNS up enough already, let's not make it worse by adding layers of TTL.