I also would like Grid to handle the state of the RC servers registered to in a lot better. As it stands, I have to restart Grid and all RC servers associated to it every time one of the RC servers gets in a bad state. Fixing this would enhance the dependability of Grid significantly.

I also think this would be a great help. When troubleshooting or developing a new grid environment I often restart individual RCs. However doing so with the current code makes it so there isn't a way to do this without shutting down the hub and all of your other RCs in the grid.

Some more helpful information: If you start up a new remote control and then stop it (stop it's process), you can't start it again because the hub still thinks it's registered, yielding:

[java] INFO: Registering new remote control...
[java] 2009-11-20 18:39:07.053::WARN: /registration-manager/register:
[java] java.lang.IllegalStateException: attempting to add a remote control that is already registered: [RemoteControlProxy localhost:4445 0/1]
[java] at com.thoughtworks.selenium.grid.hub.remotecontrol.RemoteControlProvisioner.add(RemoteControlProvisioner.java:68)
[java] at com.thoughtworks.selenium.grid.hub.remotecontrol.GlobalRemoteControlPool.register(GlobalRemoteControlPool.java:35)
[java] at com.thoughtworks.selenium.grid.hub.management.RegistrationServlet.process(RegistrationServlet.java:29)
[java] at com.thoughtworks.selenium.grid.hub.management.RegistrationManagementServlet.doPost(RegistrationManagementServlet.java:18)
[java] at javax.servlet.http.HttpServlet.service(HttpServlet.java:727)
[java] at javax.servlet.http.HttpServlet.service(HttpServlet.java:820)
[java] at org.mortbay.jetty.servlet.ServletHolder.handle(ServletHolder.java:459)
[java] at org.mortbay.jetty.servlet.ServletHandler.handle(ServletHandler.java:360)
[java] at org.mortbay.jetty.servlet.SessionHandler.handle(SessionHandler.java:231)
[java] at org.mortbay.jetty.handler.ContextHandler.handle(ContextHandler.java:629)
[java] at org.mortbay.jetty.handler.ContextHandlerCollection.handle(ContextHandlerCollection.java:149)
[java] at org.mortbay.jetty.handler.HandlerWrapper.handle(HandlerWrapper.java:141)
[java] at org.mortbay.jetty.Server.handle(Server.java:303)
[java] at org.mortbay.jetty.HttpConnection.handleRequest(HttpConnection.java:452)
[java] at org.mortbay.jetty.HttpConnection$RequestHandler.content(HttpConnection.java:735)
[java] at org.mortbay.jetty.HttpParser.parseNext(HttpParser.java:636)
[java] at org.mortbay.jetty.HttpParser.parseAvailable(HttpParser.java:209)
[java] at org.mortbay.jetty.HttpConnection.handle(HttpConnection.java:349)
[java] at org.mortbay.jetty.bio.SocketConnector$Connection.run(SocketConnector.java:217)
[java] at org.mortbay.thread.BoundedThreadPool$PoolThread.run(BoundedThreadPool.java:475)

Ah I didn't see the other issue. This is a duplicate of GRID-14 . The server simply has to handle the case of un-registration. But not only by manual shutdown, the server should be pinging the RC and un-register if it fails to respond.