On 28 April 2010 10:30, Peter Firmstone <[email protected]> wrote: > In the case I'm analyzing, if an Entry is updated during a network break, > the remote notification receives an IOException (RegistrarImpl:EventTask) > which is not retryable (see ThrowableConstant : retryable(exception) ) > IOException seem to be NOT retryable. > > Could anyone explain to me how river is supposed to handle this case and > update the Entry on remote ServiceDiscoveryManagers after the end of the > network break ?
It is not up to the Jini infrastructure to do anything in such a case. It is up to your application logic what you want to do. At least your app will know that the Entry /may/ not have been updated, and that it needs to check the state of affairs when it next gets a chance. If you were updating the Event under a Transaction, at least you will be pretty sure that the Entry has /not/ been updated, so life is a little simpler in that case. -- This email is [X]bloggable [ ]ask-first [ ]private Mike Morris http://mikro2nd.net/ EarthStuff: http://mikro2nd.net/blog/planb/ TechStuff: http://mikro2nd.net/blog/mike/ ----- A day without chillies is a day wasted ------
