I am seeing some really weird behavior when using a Registrar on Fedora Core 6. This is a problem with Jini 2.1.
I have 2 machines. The first is a Macintosh running OS-X, and the other is a PC running Fedora Core 6. I have Jini 2.1 installed on both of them.
I have edited the groups associated with the services on the Linux machine so that the groups don't match. I have found that this avoids certain other weird problems.
In any event, I run the launch-all script on both machines. On the Macintosh, the Service Browser sees the local registrar as well as the registrar that runs on Linux. The Service Browser on Linux does not see registrar on the Mac.
When the Macintosh attempts to browse the services that are registered with the Linux machine, it hangs. It remains hung until I shut down the browser on Linux. It does *not* hang when browsing the local registrar.
I have development setups on both machines, which are running a basic unicast test. In the test, I attempt to use the getRegistrar() method to get the local registrar, then I set up a second locator and use the getRegistrar() method to get the remote registrar. The locators are passed the appropriate URLs (on the Mac it gets jini://localhost for the local registrar, then the URL of the Linux machine, while on the Linux machine it gets jini://localhost for the local registrar, then the URL of the Macintosh).
On the Macintosh, the getRegistrar() method is successful in both cases. It gets the local registrar and the registrar on the Linux machine.
On the Linux machine, getRegistrar() hangs when attempting to get the registrar from the Macintosh.
I believe that the behaviors mentioned above are caused by some kind of misconfiguration on the Linux machine. I cannot, however, figure out what that misconfiguration can be.
Can someone give some insights as to why Linux is failing to properly handle Jini communications between machines?
Previous Topic
