I had these too, until I updated my DNS settings and pointed directly at the ones of my ISP (instead of my gateway)
I’m still unable to reach my intranet server unless using its IP address.
Lilian
I’ve been in discussions with the deployment team about DNS problems and one of the members of Sun’s Java SE networking team found the root cause of some of these applet-related DNS problems. They are unrelated to Java and are caused by installing MSN Messenger 7.5.
If anybody is experiencing DNS-related issues with Java, please look into this workaround and see if it’s what you’re experiencing. We’ll escalate this issue with Microsoft.
I dont use messenger, and I confirmed that my values (currentcontrolset 001 & 003) were both expand. So something else is the for me… I am available for testing 
No MSN messenger here either…
Is it possible that this UnknownHostException is related to a setup issue? Has anything on the system been changed since the last mustang version worked?
What was the last version that worked successfully?
Can we get some more information about the setup on your machine:
- is ipv6 enabled?
- Do you need a proxy to access the internet?
- is your machine setup to use DNS for name resolution?
Would it be possible to run the following code passing it the hostname that is unresolvable:
/**
- Use system configured name services
- run: java Lookup
- Use JNDI-DNS name service provider
- run: java -Dsun.net.spi.nameservice.provider.1=dns,sun Lookup
*/
"import java.net.*;
public class Lookup
{
public static void main(String[] args) throws UnknownHostException
{
if (args.length == 1) {
InetAddress addr = InetAddress.getByName(args[0]);
System.out.println(addr);
}
}
}"
What is the outcome of these 2 runs?
[quote=“Chris_Hegarty,post:25,topic:25208”]
I have 1.4, 1.5 and 1.6 installed. Only 1.6 exhibit this behaviour
1.5
- no, default xp, sp2
- no
- yes, dhcp, isp dns
java -cp bin; Lookup javagamesfactory.org
javagamesfactory.org/213.88.244.247
java -Dsun.net.spi.nameservice.provider.1=dns,sun -cp bin; Lookup javagamesfactory.org
Exception in thread "main" java.net.UnknownHostException: DNS name not found [response code 3]
at sun.net.spi.nameservice.dns.DNSNameService.resolve(DNSNameService.java:162)
at sun.net.spi.nameservice.dns.DNSNameService.lookupAllHostAddr(DNSNameService.java:315)
at java.net.InetAddress.getAddressFromNameService(Unknown Source)
at java.net.InetAddress.getAllByName0(Unknown Source)
at java.net.InetAddress.getAllByName0(Unknown Source)
at java.net.InetAddress.getAllByName(Unknown Source)
at java.net.InetAddress.getByName(Unknown Source)
at Lookup.main(Lookup.java:14)
I reran the test using 1.5 (apparently I have removed 1.4) and got the same result:
c:\Java\jdk1.5.0\bin\java.exe -cp bin; Lookup javagamesfactory.org
javagamesfactory.org/213.88.244.247
c:\Java\jdk1.5.0\bin\java.exe -Dsun.net.spi.nameservice.provider.1=dns,sun -cp bin; Look
up javagamesfactory.org
Exception in thread "main" java.net.UnknownHostException: javagamesfactory.org: DNS name not found [response code 3]
at java.net.InetAddress.getAllByName0(InetAddress.java:1128)
at java.net.InetAddress.getAllByName0(InetAddress.java:1098)
at java.net.InetAddress.getAllByName(InetAddress.java:1061)
at java.net.InetAddress.getByName(InetAddress.java:958)
at Lookup.main(Lookup.java:14)
so perhaps 1.6 defaults to the dns,sun SPI ?
http://forums.java.net/jive/thread.jspa?messageID=32852&tstart=0#32852
“No, the stack allocation of non-escaping objects is planned for the release after Mustang. Depending on the performance benefit it gives and the size and complexity of the changes required, it could be eventually backported to a future Mustang update release, but there are no current plans to do so.”
Aww, rats. Still, it is at best a minor performance enhancement, I suspect.
Cas 
With regard to the UnknownHostException being thrown for javagamesfactory.org.
Thanks to the help from matzon you can workaround this issue temporarily, until resolved by Sun, by removing the …/jre/lib/ext/dnsns.jar jarfile from your mustang install.
This is not a great workaround, but until I come up with something better at least it will allow you to test the performance of C1 which is what thie thread is about.
This UnknownHostException issue can be tracked by CR: 6363704 in Sun’s bug database.
Is there any vs C++ microbenchmarks for b59+ already?
both ibm’s jdk5 beta and jrockit5 are faster than the newest mustang server vm (not really much, but still ->)
- my connect4-ai: 2.5 million checks (sun) vs 2.8 million (bea) vs 3 million (ibm) on my machine (1.5.0_06 client crowls at 1.4 million)
- my minesweeper-solver: ibm 60 seconds, sun 62 seconds, bea 50 seconds (for a specific field)
what’s causing the difference?
Could you make the source code for these two programs available? We could probably take a look and see what HotSpot is doing.
What’s the run time of these programs? A second? Less than a second? A minute?
minesweeper:
http://www.java-gaming.org/forums/index.php?topic=12151.0
i’ll upload the connect4-stuff tomorrow.
the run time is as long as it needs to reach the maximum speed. i repeat it until it doesn’t get faster.
Why regression, as far as I understood it never ran faster using hotspot, right? Then its no regression.
I made the experience that Hotspot-Server is for my tasks (long running server apps) the best descision, mayby someone can squeeze max. 10% more out with IBM or Bea, but in overall Sun remains on performance-top with best reliability and it scales also well on larger systems.
lg Clemens
Sorry, I misread. I just spotted the JDK5 in “ibm’s jdk5 beta”, and thought this was a regression from Sun’s Java 5.
This competition amoung JVM vendors is great
(a few days later)
here’s connect4:
http://www.radixversand.ch/selfmadeproggies/KI%20Sandbox/
the comments are german. just ignore them. just run MainFrame and select the cpu level and board size. you can click anywhere (which is actually cheating), but i’ll put my trust in you here^^
set the cpu level to 11 or 12 and the board size to 7 or 8, depending on your patience.
there is some debug output that tells you what the ai thinks, and how fast it thinks.
I looked into your minesweeper and connect4 benchmarks. There are bugs in the minesweeper solver causing it to lock up on some boards. The attached field-30-30.txt works fine while the attached field-30-30-2.txt does not. Because of this I wasn’t able to test larger fields. The 30x30 fields are solved in less than 100 ms so in this configuration this benchmark is testing startup time more than anything. You’ll need to fix the problems with the solver if we’re to look into it further.
For the connect4 game, the IBM 5.0 JVM for Linux and the Sun 5.0u6 JVM (-server) both compute somewhere around 2.1 million moves per second with the same board arrangement at the same point in the game. BEA JRockit R26-5.0_04 computes around 2.3 million moves in the same situation. The problem with this benchmark is that it isn’t automatable. If you make a headless version of this game which pits two computer players against each other with the exact same board configuration and no randomization, so the results are deterministic, and it computes statistics like the peak moves per second computed, then we can use it as a benchmark and look into it further.
Basically I don’t see any performance problems with the Sun JVM as it stands with these benchmarks.