Jan Ceuleers wrote: > All, > > One of my favourite articles on the interweb is the following (about the > Netgear/UWisc debacle). > > http://www.cs.wisc.edu/~plonka/netgear-sntp > > In the list of best practice recommendations it makes about SNTP > implementations, it says: > > "SHOULD resolve the server IP address via DNS before each poll/query, so > that the pertinent DNS entries' Time-To-Live values are respected." > > I was wondering whether the readership of this forum feels that this > should also apply to NTP implementations. >
No, but not because I disagree with his recommendation but because I am intending to implement a different scheme which will have the effect of stopping the ntp server from continuing to use the IP address if it stops responding. I will be doing this as well as making use of the additional addresses returned by getaddrinfo() to make use of all the data returned. It will take a while to implement so don't expect to see it soon. In theory one should stop using an IP address once the name's TTL has expired. In practice it's harder to determine when that is because none of the standard DNS API's will give you that information. dig does that because it uses its own API's and is not dependent on the O/S's libraries. Even worse what if you specify an IP address like the Netgear equipment was doing. Do you then depend on a reverse lookup and what its TTL looks like? I would love to have getaddrinfo() to add the TTL information but I doubt that it's going to happen. Danny _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
