river-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Patricia Shanahan <p...@acm.org>
Subject Re: TaskManager progress
Date Mon, 16 Aug 2010 13:15:35 GMT
Jonathan Costers wrote:
>>
>>     [java] Test process was destroyed and returned code 1
>>     [java] com/sun/jini/test/impl/joinmanager/LeaseRenewDurRFE.td
>>     [java] Test Failed: Unexpected Exception in setup:
>> java.security.AccessControlException: origin is non-local host
>>
>>
> I had the same issue on Ubuntu.
> 
> The reason seems to be the default network configuration in Ubuntu:
> jonathan@calisto:~$ cat /etc/hosts
> 127.0.0.1    localhost
> 127.0.1.1    calisto
> 
> Notice the second line.
> 
> Now, in com.sun.jini.phoenix.LocalAccess, a check is being done on whether
> the client is a local host:
> ...
> InetAddress addr = host.getClientHost();
>     Boolean ok = (Boolean) cache.get(addr);
>     if (ok == null) {
>         try {
>         ok = Boolean.valueOf(NetworkInterface.getByInetAddress(
>                     addr) != null);
>         } catch (IOException e) {
>         ok = Boolean.FALSE;
>         }
>         cache.put(addr, ok);
>     }
>     if (!ok.booleanValue()) {
>         throw new AccessControlException("origin is non-local host");
>     }
> 
> However, this check does not work for the above configuration in Ubuntu
> (hostname associated with a loopback address, but not with 127.0.0.1).
> The check only looks at physical network interfaces to determine whether the
> client is a local host.
> 
> A fix could be to add an extra check on loopback addresses (which I tested
> and seems to work in my case):
>    ok = Boolean.valueOf(NetworkInterface.getByInetAddress(
>                     addr) != null || addr.isLoopbackAddress());
> 
> Or to change /etc/hosts (which I've also tested):
> 127.0.0.1 localhost calisto
> 

Thanks. I have tests of both these running. Assuming both also work for 
me, which is the better fix?

I'm not a networking expert, but it seems to me to be logical to treat 
any and all loop back addresses as being local, so my initial preference 
is the source code fix.

Opinions?

Patricia


Mime
View raw message