It simply is not possible to disable the cache from Java. The fixes/enhancements you mention are already planned for a future release, but didn't happen yet for various reasons.
sorry about that. On Thu, Jan 15, 2009 at 12:43 PM, Greg White <[email protected]>wrote: > All, > > I've recently been bit by some DNS caching issues and am wondering if there > is some way around them. The source seems to say 'no', but perhaps I'm > missing something. > > I'm trying to resolve hostnames using a couple of different methods: > > new InetSocketAddress(host, port) // The 'normal ' way > > and > > InetAddress[] addresses = InetAddress.getAllByName(host); // And this > way to bypass the cache in InetAddress > > I start the emulator with -dns-server pointing at my host which is running > dnsmasq set to return records with a TTL of 1 second. This allows me to > tweak the returned DNS values to simulate a DNS change. > > In each case, I'm doing this to (try) to disable the cache. I'm doing both > the generally accepted method (Security) and what PriviAction seems to be > actually doing (System). > > System.setProperty("networkaddress.cache.ttl", "0"); > Security.setProperty("networkaddress.cache.ttl", "0"); > > No luck, there is still DNS caching in place. > > Some poking around bionic reveals that there is a DNS cache in there, an > MRU cache with 128 entries and a 10 minute timeout. Sure enough, this > matches what I am seeing exactly. > > There does not seem to be any way to flush this cache. > > I can disable it by setting an environment variable, BIONIC_DNSCACHE, to > zero at process startup - if I can find a way to do that from Java. That > would be useful. > > While I am all for DNS caches, this 10 minute caching would seem to be > inappropriate for a roaming device. The problem I am trying to work around > is that when I switch from my work network to the carrier network and vice > versa, the DNS cache is not flushed. We have an internal server set to > resolve differently when you are inside the work network than when you are > external (common enough) and this switch takes 10 full minutes to take > effect after I switch networks. > > Is there some way to dump this cache short of a process restart? > > Neither of these caches honors TTL. Bad caches. > > And there are two of them, one in Java, one in C, one on top of the other, > which is very strange. One would seem to be entirely sufficient. > > Furthermore, would it not be better to centralize these caches and flush > them on a network change? It would seem that it would be more efficient, > faster and better to have all applications hitting one DNS cache, which > could then be flushed after the DNS server changed. It would be easy to > configure dnsmasq to do just this - and all of the processes' DNS caches > could be eliminated and fixed - and, presumably, this central cache could > honor TTL. It further appears that this could be accomplished with minimal > code changes by setting the correct propery in Java, the right environment > variable in C, pointing everyone at the local dnsmasq instance, and then > configuring that on network change. > > Any help which anyone can offer would be greatly appreciated. > > > > > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/android-developers?hl=en -~----------~----~----~----~------~----~------~--~---
