This forum has been archived. Please start a new discussion on GitHub.

Can you add an active-test mechanism in IceGrid?

Can you add an active-test mechanism between IceGrid node and Server, and between IceGrid Locator and IceGrid Node ? The advantage of this is:

1)The registry information of Server and IceGrid Node does not have to been saved in a file by IceGrid Node or IceGrid Locator.

2)If a Server dies(kill -9), the IceGrid Node can detect this by active-test and remove its registry infomation.

3)If IceGrid Node dies(kill -9), the Server can detect this; when the IceGrid restarts, the Server can connect it by periodically attempting, then IceGrid Node can collection all servers' information again.

3)It's the same about IceGrid Locator.

By the way, the idea above comes from VisiBroker's osagent.


  • benoit
    benoit Rennes, France
    Because of the way process monitoring is implemented, it's not possible to shutdown the node and then later have the node monitor some processes it previously forked, so adding what you are suggesting wouldn't be trivial.

    The IceGrid node will however periodically ping the IceGrid registry on startup if it can't initially reach it -- it won't require the registry to be up to start. It will also be possible to stop the registry and restart it without having to restart all the nodes.